betaFeedback

WorksPhysics and Astronomy

Emergent higher-order optical nonlinearity in self-assembled polar supertextures

Abstract

Giant optical nonlinearities enable advanced materials control and bridge exotic light-matter phenomena to next-generation photonics. However, materials exhibiting strong higher-order optical nonlinearity beyond the second order are typically rare. Here, we report the discovery of a giant and ultrafast second-harmonic generation (SHG) in self-assembled polar skyrmions under mid-infrared excitations. The transient SHG intensity exhibits a distinctive power dependence originating from higher-order nonlinearity. More generally, we show that large higher-order nonlinearity emerges across diverse domain structures in polar superlattices. Phase-field simulations demonstrate a strong and periodic local permittivity variation within each polar texture at equilibrium, and further confirm the fast polarization oscillation in polar skyrmions with mid-infrared excitations. This intense local permittivity modulation can satisfy epsilon-near-zero conditions, unlocking dramatically enhanced nonlinear optical responses without photonic engineering. Consequently, the self-assembled polar topologies with strong inherent optical nonlinearity could establish a platform for quantum information processing and high-speed optical communications.

1 Department of Physics, Boston University, 590 Commonwealth Avenue, Boston, 02215, Massachusetts, USA.

2 Department of Materials Science and Engineering, Pennsylvania State University, University Park, 16802, Pennsylvania, USA.

3 Materials Research Centre, Indian Institute of Science, 560012, Karnataka, India.

4 Department of Materials Science and Engineering, University of California, Berkeley, 94720, California, USA.

5 Keck Center for Quantum Magnetism and Department of Physics and Astronomy, Rutgers University, 136 Frelinghuysen Road, Piscataway, 08854, New Jersey, USA.

6 Present address: School of Physics Astronomy, University of Minnesota, 115 Union St. SE, Minneapolis, 55455, Minnesota, USA.

7 Department of Materials Science and Nanoengineering, Rice University, 77005, Houston, Texas, USA.

8 Department of Physics and Astronomy, Rice University, 77005, Houston, Texas, USA.

9 Department of Chemistry, Rice University, 77005, Houston, Texas, USA.

10 Rice Advanced Materials Institute, Rice University, 77005, Houston, Texas, USA.

11 Materials Sciences Division, Lawrence Berkeley National Laboratory, 94720, Berkeley, CA, USA.

12 Department of Materials Science and Engineering, University of California, Berkeley, 94720, Berkeley, CA, USA.

Introduction

Optical nonlinearity[1, 2] has revolutionized material control since the advent of lasers. Materials with strong optical nonlinearities, such as non-centrosymmetric crystals, plasmonic nanostructures, and engineered metamaterials are critical for nonlinear optical applications[3, 4, 5, 6]. Second-order nonlinear effects drive frequency doubling[7, 8, 9]. Third-order effects enable four-wave mixing including third harmonic generation[10, 11, 12]. Higher-order nonlinear optical phenomena (third order and above) in conventional materials are typically weak[13, 14], limiting their utility in integrated photonics. Materials exhibiting strong higher-order optical nonlinearities with short response times are highly desirable for optoelectronic devices that leverage multiphoton interactions and non-degenerate wave mixing processes[12]. Such devices, including photonic switches and frequency converters, are critical for applications in quantum information processing, optical remote sensing, and high-speed optical communications[15]. However, achieving these properties often requires complex and intricate material engineering. In artificially structured metamaterials and photonic crystals, the strong spatial variation of permittivity from periodically arranged materials enables the epsilon-near-zero condition, thereby greatly enhancing nonlinear optical response[16]. Alternatively, nanoscale self-assembled polar textures with large optical nonlinearity offer a new route for nanophotonics design[17].

Ferroelectric superlattices have emerged as engineered quantum materials wherein layers of ferroelectric and dielectric materials are stacked periodically at the nanoscale. In superlattices, mechanical constraints, particularly epitaxial strain from lattice mismatch, couple with electrostatic effects arising from polarization discontinuities at ferroelectric-dielectric interfaces. The coupled effects of mechanical and electrostatic boundary conditions modify the local symmetry of ferroelectric unit cells, generating stable self-organized and topologically nontrivial textures[18], including polar skyrmions[19], polar merons[20], and polar vortices[21]. These emergent polar textures host spatially varying unconventional dielectric properties[22, 23, 24], and exhibit strong light-matter interactions[25, 26] including sub-terahertz collective modes[27, 28]. The topologically nontrivial polar textures are also highly susceptible to electric fields[24, 29, 30, 31, 32, 33, 34]. Under a moderate out-of-plane electric field, the polar skyrmions can be fully polarized to a single cc-domain state[24, 32, 34], with tens of nanoseconds recovery time[24]. Despite extensive interest in these systems, little is known about their nonlinear light-matter interaction at ultrafast time scales. Potential large optical nonlinearities in polar textures remain a frontier awaiting exploration.

Here, we focus on the polar-skyrmion phase in (PbTiO3)n/(SrTiO3)n(\mathrm{PbTiO}_3)_n/(\mathrm{SrTiO}_3)_n (PTO/STO) superlattices. The impact of mid-infrared laser excitation on the second-harmonic generation (SHG) process is explored. The mid-infrared excitation induces a giant transient SHG intensity from the polar skyrmions, exhibiting a fourth-power dependence on the excitation field. In addition, the

SHG intensity oscillates at the mid-infrared driving frequency, with a dominant third-power dependence on the excitation field. This is attributed to an effective fourth-order optical response. The giant light-induced nonlinear polarization in the polar skyrmion is in stark contrast to its vanishing in-plane macroscopic polarization in equilibrium. A maximum modulation ratio of 350 is achieved with in-plane detections. This high contrast, sub-picosecond response time, and the absence of strict in-plane orientation constraints are crucial for potential applications such as photonic switches, terahertz-to-visible photon converters and terahertz detectors. More generally, we observe a strong higher-order optical nonlinearity across diverse polar textures in PTO/STO superlattices. These polar textures exhibit a strong and periodic variation of local permittivity, resemble metamaterials and photonic crystals. Investigating the large inherent higher-order optical nonlinearity in self-assembled polar textures could unlock a new paradigm for nonlinear optics[17, 35] and quantum photonic platforms[36, 37], combining the scalability of bottom-up synthesis with topological-texture-enabled light-matter interactions.

Results

Giant light-induced second-harmonic generation

In the polar-skyrmion phase, the ferroelectric polarization vectors rotate smoothly from the center of a skyrmion to its boundary, forming a three-dimensional swirl pattern with a scale of ∼10\sim10 nm[19]. At the skyrmion core, the polarization vectors align parallel to each other and are normal to the sample surface. The spatial average over the entire skyrmion leads to a strongly suppressed in-plane macroscopic ferroelectric polarization. Room-temperature SHG polarimetry measurements were conducted on (PbTiO3)16/(SrTiO3)16(\mathrm{PbTiO}_3)_{16}/(\mathrm{SrTiO}_3)_{16} superlattices grown on SrTiO3\mathrm{SrTiO}_3 (001) substrates known to possess polar skyrmions. SHG polarimetry was measured in reflection at 45∘45^\circ incidence along orthogonal directions (Figure 1a). The ss direction is parallel to the superlattice’s surface. The pp direction is parallel to the incidence plane. Overall, the equilibrium SHG intensity is weak (Figure 1b and Supporting Information Fig. S2). The pp channel signal is slightly stronger, consistent with its sensitivity to the finite out-of-plane polar response. When driving the polar skyrmion out of equilibrium with mid-infrared (MIR) laser pulses (18 THz, 230 fs), a large SHG signal is observed (Figure 1c). The maximum transient SHG intensity is about 20 times greater than the equilibrium value. Further measurements show that the maximum transient SHG intensity is insensitive to both rotation of the sample and rotation of the linear MIR polarization. Rotating the MIR polarization by 90∘90^\circ brings the ss channel transient SHG intensity up to the same level as that in Figure 1c (Supporting Information Fig. S5). Due to spatial cancellation of the winding in-plane polarization components, the equilibrium SHG intensity is an order of magnitude weaker in the ss-channel than the pp-channel (Figure S2). This results in a maximum SHG intensity modulation ratio of 350 with parallel pump and probe polarizations.

Polar skyrmion SHG polarimetry and light-induced enhancement

Figure 1. Strong light-induced SHG in the polar skyrmion. a, In equilibrium, the polar skyrmion has a net ferroelectric polarization along the sample normal. We measured SHG polarimetry using a linearly polarized 800 nm fundamental beam with varying angle ϕ\phi and two fixed detection polarizations (pp and ss). Mid-infrared excitation (16.7 µm, 18 THz) is linearly polarized and comes at normal incidence. b, A weak but finite SHG signal is seen in equilibrium. c, With the mid-infrared excitation, the SHG intensity is enhanced by a factor of 20 near ϕ=0∘\phi= 0^\circ (along one cubic edge of the sample). Here t=0t = 0 ps is defined by the maximum response.

Dramatic light-induced change of SHG

Figure S2. Dramatic light-induced change of SHG a, The SHG polarimetry was measured by varying the polarization angle ϕ\phi of the fundamental beam and with fixed detection polarization (pp or ss). b, Equilibrium SHG polarimetry with pp detection. c, Equilibrium SHG polarimetry with ss detection. The ss channel SHG intensity is one order of magnitude smaller than that of the pp channel. d, The mid-infrared (MIR) field is polarized in the SHG incidence plane. e, Equilibrium (gray) and transient (blue) SHG polarimetry with pp detection. f, Equilibrium (gray) and transient (blue) SHG polarimetry with ss detection.

Evidence of the higher-order nonlinearity

The transient SHG intensity is pronounced as compared with the equilibrium SHG from, for example, a single-domain BiFeO3\mathrm{BiFeO}_3 crystal (Figure 2b and Figure S3), a well-known ferroelectric with large polarization and large second-order optical nonlinearities. The transient SHG enhancement in the polar skyrmion sample has a lifetime comparable to the duration of the MIR pulse (Figure 2a to Figure 2c). In addition, the transient SHG intensity oscillates with time. The non-oscillatory component of the signal is fitted with a Gaussian function and defined as the displacive response. The oscillatory response is then obtained by subtracting the fitted displacive contribution from the raw data (Figure 2b, see Supporting Information S7 for details). The oscillatory response follows the MIR field (Figure 2c). A control experiment on a PbTiO3\mathrm{PbTiO}_3 film with a net out-of-plane polarization shows a similar oscillatory response but a significantly weaker displacive response (Figure 2b and Figure S4), highlighting the unique character of the superlattices. As the MIR excitation used in this work is far away from the sub-terahertz collective modes[28, 38], no other low-frequency oscillations other than that at the MIR driving frequency are observed.

Evidence of the higher-order nonlinearity

Figure 2. Evidence of the higher-order nonlinearity. a, Normalized transient SHG polarimetry as a function of pump-probe time delay. b, Time evolution of ΔIp(2ω)\Delta I_p(2\omega) at the maximum response angle. A weak oscillation can be resolved (blue open circles) after removing the displacive response (blue solid line). For comparison, we present the equilibrium SHG intensity from a [1]pc_{pc} BiFeO3_3, measured in the same setup without MIR excitation (red solid circle), together with the MIR-induced SHG intensity change of a 100-nm-thick PbTiO3_3 film with ferroelectric polarization perpendicular to the film surface (green solid line). c, An electro-optical sampling characterization of the mid-infrared field. d, For the superlattice sample, the displacive response in ΔIp(2ω)\Delta I_p(2\omega) shows a dominant quartic pump-field dependence. (Inset) In contrast, the PbTiO3_3 film shows a quadratic pump-field dependence for the displacive response. e, For the superlattice, the oscillatory response follows a dominant cubic field dependence. (Inset) In contrast, the oscillatory response of the PbTiO3_3 film scales linearly with the pump field. For both d and e, vertical error bars represent the standard deviation. The thin dashed/solid lines are guides for the eye on pure quadratic/quartic pump-field dependence for d, and pure linear/cubic pump-field dependence for e.

Figure S3. Giant transient SHG from the polar skyrmion. a, Equilibrium SHG polarimetry of a single-domain BiFeO3_3 with pp detection, measured under the same condition as the polar skyrmion, except for the absence of the mid-infrared excitation. b, The transient SHG polarimetry from the polar skyrmion.

Figure S4. PbTiO3_3 c-domain film versus skyrmion-hosting PbTiO3_3/SrTiO3_3 superlattices. The non-superlattice film exhibits only a weak MIR-induced SHG response, in stark contrast to the giant transient SHG in the skyrmion-hosting superlattice.

We next examined the pump-field dependence of the transient SHG response. When using a polynomial fit, the transient SHG intensity scales predominantly to the fourth-power of the MIR peak field (Figure 2d). The oscillatory signal shows a dominant third-order field dependence (Figure 2e). These results indicate that higher-order nonlinear optical processes are present. Under the same MIR excitation, the displacive (Figure 2d inset) and oscillatory (Figure 2e inset) responses of the PbTiO3_3 film with cc domain polarization exhibit the quadratic and linear dependencies on the pump field, respectively.

We used phase-field simulations to investigate the polarization dynamics in the light-driven state for the polar skyrmions. Figure 3a shows the stabilized distribution of ferroelectric polarization at the top, middle, and bottom of the PTO layers, exhibiting polar-skyrmion structure. The maximum light-induced change in the ferroelectric polarization is shown in Figure 3b. At time zero, the electric field of the excitation pulse reaches a maximum and is along the +x+x direction. From Figure 3d it is seen that the entire PTO layer develops a transient in-plane ferroelectric polarization following the MIR field (Figure 3c). The transient change in the magnitude of ferroelectric polarization is strongest at the top and bottom surface of the PTO layers and along the skyrmion walls. No sizable quasi-static change in the spatially averaged ferroelectric polarization is observed in the simulation (Figure 3d). Overall, the average ferroelectric polarization change exhibits a stronger oscillation along the pump-field direction (Figure 3d), while a much weaker and slower oscillation is seen for the ferroelectric polarization change along zz. The magnitude of the ferroelectric polarization change is small with respect to the maximum local electric dipole strength (Figure 3a). The rapid polarization oscillation driven by the MIR field is consistent with the oscillatory component observed in the transient SHG signal (Figure 2b).

Figure 3. Phase-field simulation of the polar-skyrmion phase. a, Equilibrium ferroelectric polarization distribution for the top, middle and bottom of the PTO layer. b, Light-induced change in ferroelectric polarization at maximum response (t=0t = 0 ps). At this instant, the pump electric field points to the right. The black arrows indicate the pump-induced change in the in-plane ferroelectric polarization. The color scale shows the percentage change in ferroelectric polarization magnitude, which is largest at the top and bottom surfaces. c, The excitation pulse used in the simulation. d, The average light-induced ferroelectric polarization change along xx (pump polarization direction), yy and zz directions.

Polarization dynamics in various polar textures

In addition to the polar skyrmions, other polar phases were examined including a mixed classical ferroelectric (a1/a2a_1/a_2) and polar-vortex phase[39], and a pure a1/a2a_1/a_2 phase[40]. All three polar phases exhibit strong light-induced SHG (Supporting Information S5) with oscillations at the MIR excitation frequency. Because nonlinear-phononic contributions may also occur in phases with a well-defined macroscopic polarization[40], we focus here on the oscillatory response and its pump-field dependence when comparing the higher-order optical response across phases. The oscillation intensity at the maximum pump fluence is shown in Figure 4a–Figure 4c. The strongest oscillation is seen in the mixed a1/a2a_1/a_2 and polar-vortex phases. The corresponding pump-field dependence of the oscillation intensity is shown in Figure 4d–Figure 4f. Cubic pump-field dependence dominates the polar-skyrmion phase, weakens in pure a1/a2a_1/a_2, while a nearly linear dependence is seen in the mixed phase.

Figure 4. Signature of the higher-order optical response in various polar phases. The top panels (a–c) show the time-domain response at the maximum pump fluence. The bottom panels (d–f) show the pump-field dependence of the oscillation intensity. The thick solid lines are fits with a mixed linear–cubic dependence (see Supporting Information Figure S9 and Table S3). The thin dashed/solid lines are guides for the eye in a pure linear/cubic pump-field dependence respectively.

We used phase-field simulations to model the ferroelectric polarization and corresponding in-plane static permittivity (ϵxx\epsilon_{xx}) of the polar-skyrmion, a1/a2a_1/a_2, and polar-vortex phases. All three polar phases exhibit a strong spatial variation in ϵxx\epsilon_{xx}. Alternating regions of large negative and positive values of ϵxx\epsilon_{xx} are observed near the domain walls for polar skyrmions (Figure 5a). In addition, ϵxx\epsilon_{xx} varies strongly along the film thickness direction, as evidenced by the side profile of the 3D plot in Figure 5a and the line-cuts in Figure 5d. In contrast, regions of positive local ϵxx\epsilon_{xx} are seen in the a1/a2a_1/a_2 phase along the domain walls (Figure 5b) with small variations along the film thickness (Figure 5e). Positive local ϵxx\epsilon_{xx} is also seen for the polar-vortex phase (Figure 5c and Figure 5f). Here we simulated a pure vortex phase for simplicity. We anticipate that additional dielectric heterogeneity may arise at the domain boundaries in the a1/a2a_1/a_2 and polar-vortex mixed phase[39], producing a broader spatial variation of ϵxx\epsilon_{xx} than in the pure a1/a2a_1/a_2 and pure polar-vortex phases. Overall, the local permittivity varies strongly and periodically throughout the PTO layer in all three polar phases, creating a natural analog to engineered metamaterials and photonic crystals.

Figure 5. Phase-field simulations of the local static permittivity distribution. Ferroelectric polarization distribution (arrows) and the local static permittivity ϵxx\epsilon_{xx} (color coded) of the PTO layer for the polar-skyrmion phase (a), the a1/a2a_1/a_2 phase (b), and the polar-vortex phase (c). Spatial variation of ϵxx\epsilon_{xx} along the xx direction and through the center of the corresponding 3D plots for the top, middle and bottom of the PTO layer for the polar-skyrmion phase (d), the a1/a2a_1/a_2 phase (e), and the vortex phase (f).

Discussion

The key findings of this work are the giant ultrafast second-harmonic generation from the polar skyrmions, and its unconventional pump-field dependence: a predominant quartic-field dependence for displacive response, and a cubic-field dependence for the oscillatory response. We now consider possible underlying mechanisms. For simplicity, throughout the discussion section the polarization PP and the electric field EE are taken to be scalar quantities[2]. The tensor contraction form can be found in the Supporting Information Section S3. We first consider nonlinear-phononic rectification[41]. If a driven infrared-active phonon couples quadratically to another structural or polar coordinate, the induced displacement can scale as the square of the pump field. A corresponding quadratic change of the SHG field can generate both quadratic and quartic contributions to the measured SHG intensity through interference with the equilibrium SHG field. Pump-field scaling alone therefore does not uniquely distinguish this mechanism. Although the nonlinear phononics effect in PbTiO3\mathrm{PbTiO_3} is predicted to be prominent[42], the lifetime of the transient state is equal to or longer than the lifetime of the driven infrared-active phonon (typically in the order of a few picoseconds).[43] In contrast, the displacive response observed here follows the ∼230\sim230 fs MIR excitation and decays within less than 1 ps (Figure 2b and Figure 2c). We therefore conclude that nonlinear-phononic rectification is unlikely to be the dominant origin of the observed ultrafast response.

In turn, we investigate nonlinear optical processes that directly match the excitation pulse duration. We start from the process which involves the third-order susceptibility χ(3)\chi^{(3)}. A short-lived SHG with oscillations at the pump frequency may arise from the electric-field-induced SHG (EFISH) process[44]. For materials with a nonzero second-order susceptibility, χ(2)\chi^{(2)}, the

SHG intensity is given by

I∝{[χ(2)+χ(3)E(ω1)]E(ω)E(ω)}2.(1)I \propto\left\{\left[\chi^{(2)}+\chi^{(3)}E(\omega_1)\right]E(\omega)E(\omega)\right\}^{2}. \tag*{(1)}

Here, ω1\omega_1 is the frequency of the pump field, while ω\omega is the fundamental frequency of the optical probe beam. This equation has three terms (see Supporting Information Section S7): (1) A term representing the SHG intensity at equilibrium. (2) A homodyned term containing [χ(3)E(ω1)]2\left[\chi^{(3)}E(\omega_1)\right]^2, which includes a displacive term and an oscillatory term (the SHG intensity oscillates at 2ω12\omega_1, filtered out in our detection scheme). Both intensities scale quadratically with the pump-field amplitude. (3) A heterodyned term proportional to χ(2)χ(3)E(ω1)\chi^{(2)}\chi^{(3)}E(\omega_1), which oscillates at ω1\omega_1 with an intensity that scales linearly with the pump-field amplitude. The control experiment on a PbTiO3_3 film exhibits precisely this scaling behavior (Figure 2d and 2e insets). In the skyrmion case, the transient SHG intensity oscillation at the pump frequency follows a leading cubic pump-field dependence (Figure 2e), and the displacive response shows a predominant quartic pump-field dependence (Figure 2d). Therefore, higher-order nonlinear process needs to be considered.

Continuing on, the process involving the fourth-order optical susceptibility χ(4)\chi^{(4)} is examined. In this case, the second-harmonic generation intensity is given by

I∝{[χ(2)+χ(3)E(ω1)+χ(4)E(ω1)E(ω1)]E(ω)E(ω)}2.(2)I \propto\left\{\left[\chi^{(2)}+\chi^{(3)}E(\omega_1)+\chi^{(4)}E(\omega_1)E(\omega_1)\right]E(\omega)E(\omega)\right\}^{2}. \tag*{(2)}

which contains six terms (see Supporting Information Section S7). Among these, three terms contribute to the displacive response: one homodyned term with [χ(3)E(ω1)]2\left[\chi^{(3)}E(\omega_1)\right]^2 and one heterodyned term with χ(2)χ(4)E(ω1)E(ω1)\chi^{(2)}\chi^{(4)}E(\omega_1)E(\omega_1) which follows a quadratic pump-field dependence, and one homodyned term containing [χ(4)E(ω1)E(ω1)]2\left[\chi^{(4)}E(\omega_1)E(\omega_1)\right]^2 which follows a quartic pump-field dependence. This agrees with observations as shown in Figure 2d. Two heterodyned terms create oscillations at the pump frequency: the term containing χ(2)χ(3)E(ω1)\chi^{(2)}\chi^{(3)}E(\omega_1) exhibits a linear pump-field dependence, and the term containing χ(3)χ(4)E(ω1)E(ω1)E(ω1)\chi^{(3)}\chi^{(4)}E(\omega_1)E(\omega_1)E(\omega_1) follows a cubic pump-field dependence. This agrees with observation as shown in Figure 2e. The above mixed processes also produce oscillations at nω1n\omega_1 (n=2,3,4n=2,3,4), which are beyond the detection bandwidth in the measurement.

At the microscopic level, the term containing χ(4)\chi^{(4)} in Eq. (26) may arise from a cascaded process involving lower-order nonlinearities. For materials with a nonzero χ(2)\chi^{(2)}, an optical pump field E(ω1)E(\omega_1) can lead to transient polarization through the mechanism of optical rectification[45], described by

POR(0)∝χ(2)E(ω1)E(−ω1).(3)P_{\mathrm{OR}}(0) \propto\chi^{(2)}E(\omega_1)E(-\omega_1). \tag*{(3)}

The associated rectified electric field, EOR(0)E_{\mathrm{OR}}(0), can subsequently enter a third-order interaction,

P(2ω)∝χ(3)E(ω)E(ω)EOR(0).(4)P(2\omega) \propto\chi^{(3)}E(\omega)E(\omega)E_{\mathrm{OR}}(0). \tag*{(4)}

Because EOR(0)∝E2(ω1)E_{\mathrm{OR}}(0) \propto E^2(\omega_1), this cascaded pathway has the same pump-field dependence as a direct fourth-order process. To further investigate a potential optical rectification effect, we conducted terahertz (THz) emission measurements under MIR excitation, which probes the radiative optical-rectification response. No sample-generated far-field THz emission was resolved under our measurement conditions, and the sample exhibited negligible THz absorption (Figure S7). The THz-emission measurement constrains a radiative macroscopic rectification component in the measured geometry but does not exclude local or predominantly out-of-plane rectification. Within the effective 4mm4\mathit{mm} description, the strong spatial cancellation of the winding in-plane response coexists with a finite out-of-plane polar component in the skyrmion phase. An in-plane MIR field then generates a rectified polarization along the sample normal zz (Supporting Information Section S6), consistent with the absence of THz emission probed along the MIR propagation direction. Thus, both direct fourth-order and cascaded contributions can be the underlying mechanisms for our observation.

Figure S7. Characterization of the excitation pulse and potential optical rectification. a, Electro-optical (EO) sampling in a 0.2-mm thick GaSe to characterize the MIR excitation. Either a gold mirror or the polar skyrmion sample was placed at the sample position. The MIR incident angle is 22.5°. EO sampling of the gold-mirror-reflected MIR in the time domain (b) and in the frequency domain (c). EO sampling result of the MIR after interacting with the polar skyrmion: time-domain result (d) and normalized FFT amplitude (e). f, THz emission setup. A MIR filter was placed either after the sample (pos.1) or before the sample (pos.2). g, A weak THz signal was detected when using the gold mirror. h, Normalized FFT amplitude. i, A weak THz signal nearly identical to the gold-mirror case was detected from the polar skyrmion, indicating the absence of additional THz emission from the polar skyrmion. j, Normalized FFT amplitude. k, A near unity FFT ratio of the THz signal is seen, with the filter behind the sample versus in front of the sample. Therefore, the THz signal is from the MIR generation, but not from optical rectification of the MIR. All measurements were done in air without purging. The strong oscillations in the time domain after the main THz pulse and the features in the frequency domain are due to water absorption.

More generally, the transient SHG signal is the far-field result of a coherent superposition of local nonlinear polarization sources across the textured medium. Accordingly, the χ(3)\chi^{(3)} and χ(4)\chi^{(4)} terms discussed here should be regarded as effective susceptibilities, consistent with the concept of an effective nonlinear optical susceptibility commonly used in photonic crystals and metamaterials. The third- and fourth-order optical responses therefore reflect both the material properties of the polar texture and the strong-field mixing conditions. As shown in Supporting Information Figure S9–Figure S9 and Table S3, the strong MIR field is important for making the effective fourth-order contribution visible. Nevertheless, the PbTiO3_3 control film under the same MIR excitation is dominated by the lower-order field dependence, showing that field strength alone does not account for the superlattice behavior. The fits further show that the degree of deviation from a purely linear oscillatory response varies across the polar phases (Figure S9–Figure S9 and Table S3), consistent with a texture-dependent nonlinear response. Although the skyrmion phase has the weakest absolute oscillatory signal among the polar phases (Figure 4a–Figure 4c), its equilibrium SHG intensity is about 10–20 times smaller (Figure S6), resulting in the strongest normalized oscillation intensity (Figure S9–Figure S9).

Figure S6. Polar texture dependence. Equilibrium versus transient SHG for the polar skyrmion, the a1/a2a_1/a_2 sample, and the mixed phase sample. Here the MIR polarization is horizontal (along 0∘0^\circ). For all samples, the transient responses at t=0t = 0 ps are shown.

polar textureab
polar skyrmion0.0342050.003506
a1/a2a_{1}/a_{2}0.0829820.002556
a1/a2a_{1}/a_{2} + polar vortex0.1376050.001169

Table S3. Fitting coefficients for ΔIosc=aEMIR+bEMIR3\Delta I^{\mathrm{osc}} = aE_{\mathrm{MIR}} + bE_{\mathrm{MIR}}^{3}.

Figure S9. Evidence of the χ(3)\chi^{(3)} and χ(4)\chi^{(4)} contributions to the oscillatory response. The transient SHG intensity oscillates at the pump frequency, with a combined linear and cubic pump-field dependence. This figure shows the same data as Figure 4 in the main text, with the individual contributions of the linear (solid line) and cubic (dashed line) terms to the oscillatory response highlighted. Fitting coefficients for ΔIosc=aEMIR+bEMIR3\Delta I^{\mathrm{osc}} = aE_{\mathrm{MIR}} + bE_{\mathrm{MIR}}^{3} are shown in Table S3.

We finally consider material properties that may contribute to the enhanced effective nonlinear response. Previous studies report strong spatial variations of the static permittivity in polar skyrmions, including negative values near the skyrmion walls, together with a high susceptibility to external electric fields[24, 34]. These observations primarily concern the out-of-plane dielectric response, whereas the MIR excitation used here is polarized within the plane of the skyrmion-containing layers. Negative local static permittivity has been linked to a flat Landau free-energy landscape[23, 24]. Although the MIR pump lies near a PTO optical-phonon frequency[46], the sub-picosecond response discussed above disfavors a long-lived phonon-driven displaced state[40]. We notice that all three polar phases exhibit strong and periodic spatial variations of the calculated static permittivity. Among three phases, the polar-skyrmion phase has alternating regions of large negative and large positive permittivity, which may lead to the largest permittivity renormalization. Although the impact of polar topology on the dielectric dispersion is beyond the scope of this research, a largely suppressed static permittivity could reshape the dielectric dispersion and fulfill the epsilon-near-zero condition, which enhances the effective field within the material. For example, in metamaterials and photonic crystals, epsilon-near-zero conditions are achieved by mechanically engineering periodic nanostructures with a designed spatial variation of permittivity. This is the key mechanism that leads to dramatically enhanced nonlinear optical response[16, 47, 48, 49]. The advantage here is that the spatial variation of negative and positive permittivity regions is inherent to the polar texture. As the negative permittivity is seen for various polar phases, depending on applied field direction[23, 24], polar textures could potentially serve as next-generation nanophotonic materials, where large optical nonlinearities emerge from self-assembled polar topologies. Establishing a quantitative connection between the calculated static permittivity patterns and the observed nonlinear enhancement will require frequency-dependent dielectric measurements or calculations together with a local-field analysis. We therefore regard the nanoscale dielectric heterogeneity as a possible materials-level contributor to the effective nonlinear response rather than as an identified microscopic mechanism. Clarifying this connection may provide a route to using the self-organized dielectric landscape of polar textures to engineer nonlinear optical responses without externally patterned photonic structures.

Conclusions

To conclude, this work has demonstrated pronounced nonlinear polarization in the polar skyrmions. Ambient temperature operation, ultrafast response time, and the absence of in-plane orientation constraints for the polar skyrmions are significant benefits as mid-infrared sensors. In addition, the large contrast in orthogonal detection channels allows polarization-selective sensing. Furthermore, harnessing the higher-order nonlinearity may unlock photonics applications that leverage multiphoton interactions and non-degenerate wave mixing processes such as terahertz-to-visible photon converters. More generally, a strong higher-order optical nonlinearity is observed in various polar textures, including the a1/a2a_1/a_2 and the mixed polar vortex phases. All polar textures exhibit a periodic modulation of permittivity, resembling nano-engineered metamaterials and photonic crystals. The inherent strong nonlinearity from the rich polar topologies could uniquely mediate nonlinear light-matter interactions, which positions polar textures as novel nanophotonic materials for quantum information processing and high-speed optical communications.

Methods

Sample specifications

The polar skyrmion sample is a 100-nm-thick [(PbTiO3)16/(SrTiO3)16]8[(\mathrm{PbTiO}_3)_{16}/(\mathrm{SrTiO}_3)_{16}]_8 superlattice grown on SrTiO3\mathrm{SrTiO}_3 (001) substrate. The a1/a2a_1/a_2 phase sample is a [(PbTiO3)16/(SrTiO3)16]8[(\mathrm{PbTiO}_3)_{16}/(\mathrm{SrTiO}_3)_{16}]_8 superlattice on TbScO3\mathrm{TbScO}_3 (110) substrate. The mixed phase sample is grown on DyScO3\mathrm{DyScO}_3 (110) substrate with a 3-unit-cell SrRuO3\mathrm{SrRuO}_3 buffer layer followed by [(PbTiO3)16/(SrTiO3)16]7(PbTiO3)16(SrTiO3)6[(\mathrm{PbTiO}_3)_{16}/(\mathrm{SrTiO}_3)_{16}]_7(\mathrm{PbTiO}_3)_{16}(\mathrm{SrTiO}_3)_6. Samples were fabricated using pulsed laser deposition with a KrF excimer laser as described previously in Refs. [19, 21]. Films were deposited from Pb1.2TiO3\mathrm{Pb}_{1.2}\mathrm{TiO}_3 and SrTiO3\mathrm{SrTiO}_3 targets at 700 ∘C700\,^\circ\mathrm{C} in 100 mTorr O2\mathrm{O}_2 pressure with a fluence of 2 J cm−22\ \mathrm{J}\ \mathrm{cm}^{-2} on (110)-oriented DyScO3_3, (110)-oriented TbScO3_3, and (001)-oriented SrTiO3_3 substrates for the mixed phase, a1/a2a_1/a_2, and skyrmion samples respectively. Layer thicknesses were continuously monitored in situ using reflection high-energy electron diffraction (RHEED) along the [100] direction and confirmed ex situ using lab-based x-ray diffraction (Supporting Information § 12) or synchrotron-based three-dimensional-reciprocal space mapping as reported in Refs. [19, 25, 28]. The PbTiO3_3 film used as the control sample was grown on SrTiO3_3 (001) substrates under the same growth conditions used for the superlattices. The film thickness was monitored in situ by RHEED during deposition. The PbTiO3_3 film exhibits a predominantly out-of-plane polarization.

The BiFeO3_3 is predominantly a single domain single crystal, cut along the pseudocubic [1] direction. The details of the growth and sample characterization can be found in Ref. [50].

MIR pump-SHG probe setup

Room-temperature SHG measurements were performed in reflection geometry with 45∘45^\circ incidence. The fundamental beam for SHG has a center wavelength of 800 nm, with 6 nJ energy and 40 fs pulse duration. A half-wave plate was used to rotate the polarization of the fundamental beam. The probe spot size was 78 micron in diameter on the sample. The SHG beam was collimated at 40 mm after the sample to reduce spatial separation of χ(2)\chi^{(2)} and higher-order nonlinear processes. A Glan-Taylor prism was used to select the desired polarization for detection in a photo-multiplier tube. The SHG bandwidth is 10 nm (full width at half maximum, FWHM), as measured by the SHG process in a BBO crystal. A 400 nm bandpass filter with a 10 nm FWHM was used for SHG detection. The photomultiplier records the SHG intensity integrated over the spectral window transmitted by the filter. Due to the finite SHG bandwidth, contributions from higher-order nonlinear processes may enter the detected signal depending on their spectral overlap with the filter transmission. The relevant frequency components are summarized in Supporting Information Table S2.

MIR pump pulses were generated by optical parametric amplification and subsequent difference frequency generation in a GaSe crystal. The MIR spot size on sample was 176 micron in diameter, with a maximum 11 mJ cm−211\ \mathrm{mJ}\ \mathrm{cm}^{-2} fluence at the longest wavelength (16.7 micron/18 THz). The MIR pump was at normal incidence on the sample.

Phase-field simulations

Phase-field simulations [51, 52] were conducted to investigate the formation of the polar-skyrmion phase, the a1/a2a_1/a_2 phase and the polar-vortex phase in ferroelectric superlattices. The system was modeled using a three-dimensional mesh (289 × 250 × 32 for the polar-skyrmion phase, 64 × 64 × 32 for the a1/a2a_1/a_2 phase, and 128 × 128 × 32 for the polar-vortex phase) with a grid spacing of 0.4 nm. The polarization components PiP_i (where i∈{1,2,3}i \in\{1, 2, 3\}) were used as the order parameters and evolved according to the polarization dynamic equation:

μ∂2Pi∂t2+γ∂Pi∂t=−δFδPi(5)\mu\frac{\partial^{2}P_i}{\partial t^{2}}+\gamma\frac{\partial P_i}{\partial t}=-\frac{\delta F}{\delta P_i} \tag*{(5)}

where μ\mu stands for the effective inertial coefficient of polarization, γ\gamma is the damping constant, and F=∫f dVF = \int f\,dV represents the total free energy over the simulated volume. The free energy includes contributions from Landau energy, electrostatic energy, elastic energy, and gradient energy:

F=∫(fLandau+felastic+felectric+fgradient)dV.(6)F=\int\left(f_{\mathrm{Landau}}+f_{\mathrm{elastic}}+f_{\mathrm{electric}}+f_{\mathrm{gradient}}\right)dV. \tag*{(6)}

Further details on the energy functions, numerical implementation, and simulation parameters can be found in the Supporting Information. Periodic boundary conditions were applied in all three spatial directions. The film was composed of alternating 16-unit-cell layers of PbTiO3_3 and SrTiO3_3 by changing the material coefficients using a sharp interface description. Each layer of the superlattice is coherently strained to the SrTiO3_3 substrate which subjects the SrTiO3_3 layer to 0% strain, and the PbTiO3_3 layer to −1.34%-1.34\% strain for the polar-skyrmion phase, and no strain for the a1/a2a_1/a_2 phase. For the polar-vortex phase, the strain is anisotropic. The strain components are ϵxx=1.2%\epsilon_{xx}=1.2\% and ϵyy=1.05%\epsilon_{yy}=1.05\% for SrTiO3_3, and ϵxx=−0.12%\epsilon_{xx}=-0.12\% and ϵyy=−0.275%\epsilon_{yy}=-0.275\% for PbTiO3_3. The local static permittivity ϵii\epsilon_{ii} (i=x,y,zi=x,y,z) is evaluated by applying a small dc electric field, and using ϵ0ϵii=Kij+ΔPiΔEi\epsilon_0\epsilon_{ii}=K_{ij}+\frac{\Delta P_i}{\Delta E_i}, in which ϵ0\epsilon_0 is the permittivity of free space, KijK_{ij} is the dielectric permittivity tensor components, ΔPi\Delta P_i and ΔEi\Delta E_i are the local ferroelectric polarization change and local electric field change, respectively[24]. Because this quantity is evaluated using a small dc electric-field perturbation, it characterizes the static dielectric response of the phase-field model and should not be interpreted as the frequency-dependent complex permittivity at the optical frequencies used in the experiment.

To study the light-induced dynamics of skyrmions, an ultrashort laser pulse with a peak electric field E0=6.14E_0=6.14 MV/cm and a Gaussian temporal profile (FWHM = 230 fs) was applied. The light–matter interaction was introduced as a time-dependent uniform electric field. Initial polarization conditions included random noise of magnitude 0.0001 C/m2^2 to mimic finite-temperature fluctuations during annealing.

Supporting Information

The Supporting Information is available online. Sample characterization; Equilibrium and pump-induced SHG; SHG polarimetry fit; Pump polarization dependence; Polar texture dependence; Absence of THz emission; Heterodyned and homodyned SHG components; Pump wavelength dependence; Phase-field simulation.

Acknowledgments

We acknowledge Andrea Cavalleri, Aaron Lindenberg, Alexander von Hoegen and Albert Liu for discussions. D.A.B.L., D.S.G. and W.H. acknowledge support from the U.S. Department of Energy, Office of Basic Energy Sciences Early Career Research Program under Award Number DE-SC0021305. D.S.G. and W. H. additionally acknowledge support from the U.S. Department of Energy, Basic Energy Science, grant DE-SC0026026. X.L., V.A.S., L.W.M., L-Q.C. and V.G. acknowledge support from the U.S. Department of Energy, Basic Energy Science, grant DE-SC0012375. A.S., L.W.M. and R.R. additionally acknowledge the support of the Army Research Office under the ETHOS MURI via cooperative agreement W911NF-21-2-0162. A.R. acknowledges the support of the NSF Graduate Research Fellowship Program under Grant No. DGE1255832. S.D. acknowledges the Scheme for Transformational and Advanced Research in Sciences (MoESTARS/STARS-2/2023-0048), DST-FIST (SR/FST/ET-II/2023/1233 (C), and the Indian Institute of Science start up grant for financial support. M.T. would like to acknowledge IISc doctoral Fellowship. This work was supported by the W. M. Keck foundation grant to the Keck Center for Quantum Magnetism at Rutgers University. The computational work reported on in this paper was performed on the Shared Computing Cluster which is administered by Boston University’s Research Computing Services.

Author contributions

W.H. conceived the project, designed and performed the pump-probe measurements. D.A.B.L. fit the SHG polarimetry data with support from X.L. and V.G.. D.A.B.L. and D.S.G. analyzed the SHG fluence dependence. D.A.B.L. performed the phase-field simulations, with support from A.R. and L.-Q.C.. X.X. and S-W.C. grew the BiFeO3\mathrm{BiFeO}_{3} sample. M.T. and S.D. synthesized the PbTiO3\mathrm{PbTiO}_{3} film. S.D., P.M., A.S., L.M. and R.R. synthesized the PTO/STO superlattices. S.D., P.M., and A.S. conducted lab-based x-ray characterizations. V.A.S. and V.G. conducted synchrotron-based three-dimensional-reciprocal space mapping. W.H. drafted the main manuscript, and together with D.A.B.L. drafted the Supporting Information; all authors contributed to reviewing and editing the final paper.

Competing interests

The authors declare that they have no competing interests.

Data Availability

All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supporting Information.

Sample characterization

Superlattice (SL) samples used in this study were characterized by either lab-based x-ray diffraction or synchrotron-based three-dimensional-reciprocal space mapping, as shown in Figure S1.

X-ray characterization of three superlattice samples, showing polar skyrmions, a1/a2 domain pattern, and mixed a1/a2 + vortex domain measurements

Figure S1. X-ray characterization of three superlattice samples.

For the polar skyrmion sample, the left panel of Figure S1 is the reciprocal space map (RSM) in the Q[100]Q_{[100]}–Q[001]Q_{[001]} plane around the SrTiO3_3 (STO) (002) reflection. The RSM shows the STO (002) reflection, along with several orders of (PbTiO3_3)n_n/(SrTiO3_3)n_n SL reflections, labeled as ±l\pm l. The superlattice peaks are separated by ΔQ[001]=2π/d\Delta Q_{[001]} = 2\pi/d. Here dd is the thickness of the total repeating units. The superlattice peaks are labeled such that the SL (002) reflection is l=0l=0. Positive and negative integers are assigned to the peaks at higher and lower Q[001]Q_{[001]}, respectively. The right panel of Figure S1 is a line cut along Q[100]Q_{[100]} direction across SL l=−1l=-1 reflection (red dashed line in Figure S1 left panel). The two broad satellite peaks around the SL l=−1l=-1 reflections are the domain diffuse scattering from the skyrmion periodicity.

For the a1/a2a_1/a_2 sample, synchrotron-based RSM is shown in Figure S1. The left panel is the RSM in the Q[001]Q_{[001]}–ΔQ[110]\Delta Q_{[110]} plane around the TbScO3_3 (TSO) (103) reflection. Here ΔQ[110]\Delta Q_{[110]} is defined as Q[110]−Q[110]Q_{[110]} - Q_{[110]}(TSO (002)). The RSM contains the TSO (103) reflection, along with multiple orders of (PbTiO3_3)n_n/(SrTiO3_3)n_n SL reflections, labeled as ±l\pm l. Several orders of the satellite reflections arising from the a1/a2a_1/a_2 domain patterns are observed around each SL reflection. The (103) reflection was chosen because the a1/a2a_1/a_2 domain satellite peaks are not visible in the Q[100]Q_{[100]}–Q[001]Q_{[001]} plane. The right panel of Figure S1 is a line cut along the ΔQ[110]\Delta Q_{[110]} direction (red dashed line in Figure S1 left panel). Two orders of satellite peaks, labeled as ±m\pm m, are associated with the a1/a2a_1/a_2 domain modulation.

For the a1/a2a_1/a_2 and vortex mixed phase sample, the left panel of Figure S1 is the RSM in the Q[001]Q_{[001]}–Q[100]Q_{[100]} plane around the DyScO3_3 (DSO) (002) reflection. The RSM shows the DSO (002) reflection, along with multiple orders of (PbTiO3_3)n_n/(SrTiO3_3)n_n SL reflections. All SL reflections split, suggesting a mixed phase domain pattern. The split peaks are labeled as FE ±l\pm l (from the a1/a2a_1/a_2 domains), and V ±l\pm l (from the vortex domains). Several orders of the satellite reflections, arising from the vortex domain periodicity, are observed around each SL reflection. However, satellite peaks around the FE peaks are absent, since a1/a2a_1/a_2 domain satellites are not visible in the Q[100]Q_{[100]}–Q[001]Q_{[001]} plane. The right panel is a line cut around vortex l=−1l=-1 reflection along the Q[100]Q_{[100]} direction (red dashed line in Figure S1 left panel). This line cut reveals two orders of satellite peaks, labeled as ±m\pm m, associated with the vortex domain modulation. The peaks are absent in the line cut around the FE l=−1l=-1 reflection (blue dashed line in Figure S1 left panel).

Equilibrium and pump-induced SHG

In a simplified picture, the winding in-plane polarization components of the skyrmion largely cancel upon spatial averaging, while a finite out-of-plane

polar component remains. This strongly suppresses the macroscopic in-plane second-harmonic generation (SHG) response, while the finite out-of-plane polar component can still contribute to the pp-polarized equilibrium SHG signal at visible-light wavelengths. We conducted room-temperature SHG polarimetry measurements on a (PbTiO3)16/(SrTiO3)16(\mathrm{PbTiO}_{3})_{16}/(\mathrm{SrTiO}_{3})_{16} polar skyrmion sample in reflection geometry (Figure S2). The 45-degree incident angle allows us to optically access both the in- and out-of-plane information. The SHG intensity was measured along two orthogonal directions. The ss direction is perpendicular to the plane of incidence, and parallel to the superlattice’s surface. The pp direction is parallel to the plane of incidence. Overall, the equilibrium SHG intensity is small, with the pp channel signal about 10 times larger than that of the ss channel (Figure S2 vs. Figure S2), consistent with the sensitivity of the pp channel to the finite out-of-plane polar response. In the transient state, the light-induced SHG intensity is about 20 times larger than the equilibrium value for both channels (Figure S2 and Figure S2).

To demonstrate the giant MIR-induced transient SHG from the polar skyrmion sample, here we compare it with the equilibrium SHG polarimetry measured in the same setup (without MIR pumping) on a single-domain (001)pc_{\mathrm{pc}} BiFeO3_3 (Figure S3). We also compare the transient response from a 100-nm-thick PbTiO3_3 film with out-of-plane ferroelectric polarization. With the same MIR excitation, the non-superlattice PbTiO3_3 film exhibits a substantially weaker SHG intensity change than the skyrmion-hosting superlattice (Figure S4).

SHG polarimetry fit

In this section, we describe the symmetry model used to calculate the equilibrium and MIR-induced SHG response of the skyrmion phase. At room temperature, PbTiO3_3 (PTO) is in the tetragonal phase with 4mm4mm symmetry. We define a global sample coordinate system (X,Y,Z)(X,Y,Z), where ZZ is normal to the film surface, XX lies in the SHG plane of incidence, and YY is perpendicular to the plane of incidence. Because the skyrmion is a spatially varying polarization texture rather than a homogeneous domain, we distinguish this global sample frame from the local crystal axes (X1,X2,X3)(X_1,X_2,X_3), where X3X_3 is chosen along the local polar axis and X1X_1 and X2X_2 span the plane perpendicular to X3X_3.

To describe the measured macroscopic SHG response, we approximate the skyrmion phase by an effective single-domain 4mm4mm response with its polar axis along the sample normal ZZ. This is an effective symmetry description of the macroscopic optical response rather than a microscopic description of the full skyrmion texture. This approximation captures the leading macroscopic symmetry because the in-plane polarization components associated with the winding texture largely cancel upon spatial averaging, leaving an effective response predominantly characterized by the out-of-plane polar axis. We therefore choose X1∥XX_1 \parallel X, X2∥YX_2 \parallel Y, and X3∥ZX_3 \parallel Z. In this effective crystal frame, the second-order nonlinear polarization can be written as

P(2)=(P1(2)P2(2)P3(2))=(0000d150000d1500d31d31d33000)(E12E22E322E2E32E1E32E1E2).(7)\mathbf{P}^{(2)} = \begin{pmatrix} P_1^{(2)} \\ P_2^{(2)} \\ P_3^{(2)} \end{pmatrix} = \begin{pmatrix} 0 & 0 & 0 & 0 & d_{15} & 0 \\ 0 & 0 & 0 & d_{15} & 0 & 0 \\ d_{31} & d_{31} & d_{33} & 0 & 0 & 0 \end{pmatrix} \begin{pmatrix} E_1^2 \\ E_2^2 \\ E_3^2 \\ 2E_2E_3 \\ 2E_1E_3 \\ 2E_1E_2 \end{pmatrix}. \tag*{(7)}

Here the indices 1, 2, and 3 refer to the local crystal axes X1X_1, X2X_2, and X3X_3, respectively, EiE_i are the components of the fundamental optical field in this frame, and Pi(2)P_i^{(2)} are the corresponding components of the nonlinear polarization at the second-harmonic frequency. The coefficients d15d_{15}, d31d_{31}, and d33d_{33} denote the independent nonzero components of the second-order susceptibility χ(2)\chi^{(2)} in contracted notation for 4mm4mm symmetry.

Fresnel coefficients

To relate the nonlinear polarization inside the film to the detected SHG field, we account for reflection, transmission, and multiple reflections using Fresnel/local-field factors [53]. We represent the optical structure by an effective three-medium model in which media 1, 2, and 3 correspond to air, PTO, and the substrate, respectively. Since the XZXZ plane is the plane of incidence, the XX and ZZ components belong to the pp-polarized field, whereas the YY component is ss polarized.

At the fundamental optical frequency ω\omega, the Fresnel/local-field factors are

LXX(ω)=t12p1+r12pr23pe2iβ(ω)(1−r23pe2iβ(ω))cos⁡θ2cos⁡θ1,LYY(ω)=t12s1+r12sr23se2iβ(ω)(1+r23se2iβ(ω)),LZZ(ω)=t12p1+r12pr23pe2iβ(ω)(1+r23pe2iβ(ω))n1(ω)n2(ω)ninterface2(ω).(8)\begin{aligned} L_{XX}(\omega) &= \frac{t_{12}^{p}}{1+r_{12}^{p}r_{23}^{p}e^{2i\beta(\omega)}}\left(1-r_{23}^{p}e^{2i\beta(\omega)}\right)\frac{\cos\theta_{2}}{\cos\theta_{1}},\\ L_{YY}(\omega) &= \frac{t_{12}^{s}}{1+r_{12}^{s}r_{23}^{s}e^{2i\beta(\omega)}}\left(1+r_{23}^{s}e^{2i\beta(\omega)}\right),\\ L_{ZZ}(\omega) &= \frac{t_{12}^{p}}{1+r_{12}^{p}r_{23}^{p}e^{2i\beta(\omega)}}\left(1+r_{23}^{p}e^{2i\beta(\omega)}\right)\frac{n_{1}(\omega)n_{2}(\omega)}{n_{\mathrm{interface}}^{2}(\omega)}. \tag*{(8)} \end{aligned}

The same expressions are evaluated at 2ω2\omega for the outgoing SHG field. Here nin_i and θi\theta_i are the refractive index and propagation angle in medium ii, respectively, evaluated at the relevant optical frequency. The phase acquired by the wave while propagating through medium 2 is β(ω)=(2π/λω)n2(ω)dcos⁡θ2\beta(\omega)=(2\pi/\lambda_{\omega})n_{2}(\omega)d\cos\theta_{2}, where λω\lambda_{\omega} is the vacuum wavelength corresponding to the optical frequency ω\omega and dd is the effective PTO thickness represented by medium 2 in this three-medium model. Following Ref. [53], with medium 2 taken as the nonlinear medium, ninterface=n2n_{\mathrm{interface}}=n_{2} for the incoming fundamental field and ninterface=n1n_{\mathrm{interface}}=n_{1} for the outgoing SHG field. The linear Fresnel reflection and transmission coefficients are

rijp=njcos⁡θi−nicos⁡θjnjcos⁡θi+nicos⁡θj,rijs=nicos⁡θi−njcos⁡θjnicos⁡θi+njcos⁡θj,tijp=2nicos⁡θinjcos⁡θi+nicos⁡θj,tijs=2nicos⁡θinicos⁡θi+njcos⁡θj.(9)\begin{aligned} r_{ij}^{p} &= \frac{n_j\cos\theta_i-n_i\cos\theta_j}{n_j\cos\theta_i+n_i\cos\theta_j}, & r_{ij}^{s} &= \frac{n_i\cos\theta_i-n_j\cos\theta_j}{n_i\cos\theta_i+n_j\cos\theta_j},\\ t_{ij}^{p} &= \frac{2n_i\cos\theta_i}{n_j\cos\theta_i+n_i\cos\theta_j}, & t_{ij}^{s} &= \frac{2n_i\cos\theta_i}{n_i\cos\theta_i+n_j\cos\theta_j}. \tag*{(9)} \end{aligned}

The input and output Fresnel factors are included by replacing each susceptibility component with its Fresnel-dressed effective value, χijk(2)→Lii(2ω)Ljj(ω)Lkk(ω)χijk(2)\chi_{ijk}^{(2)}\rightarrow L_{ii}^{(2\omega)}L_{jj}^{(\omega)}L_{kk}^{(\omega)}\chi_{ijk}^{(2)}, where ω\omega corresponds to the 800 nm probe.

For the 45∘45^\circ incidence geometry, the fundamental probe field expressed in the sample frame has components EX=E0cos⁡ϕ/2E_X=E_0\cos\phi/\sqrt{2}, EY=E0sin⁡ϕE_Y=E_0\sin\phi, and EZ=E0cos⁡ϕ/2E_Z=E_0\cos\phi/\sqrt{2}, where ϕ=0∘\phi=0^\circ corresponds to probe polarization in the SHG plane of incidence. Since the local crystal axes of the effective skyrmion model are aligned with the sample axes, substituting these field components into Eq. (7) and including the input and output Fresnel factors gives the equilibrium SHG nonlinear polarization

Psamp(2)=E02(LXX(2ω)LXX(ω)LZZ(ω)d15cos⁡2ϕ12LYY(2ω)LYY(ω)LZZ(ω)d15sin⁡(2ϕ)12LZZ(2ω)[((LZZ(ω))2d33+(LXX(ω))2d31)cos⁡2ϕ+2(LYY(ω))2d31sin⁡2ϕ]).(10)\mathbf{P}_{\mathrm{samp}}^{(2)}=E_0^2 \begin{pmatrix} L_{XX}^{(2\omega)}L_{XX}^{(\omega)}L_{ZZ}^{(\omega)}d_{15}\cos^2\phi\\ \frac{1}{\sqrt{2}}L_{YY}^{(2\omega)}L_{YY}^{(\omega)}L_{ZZ}^{(\omega)}d_{15}\sin(2\phi)\\ \frac{1}{2}L_{ZZ}^{(2\omega)}\left[\left(\left(L_{ZZ}^{(\omega)}\right)^2d_{33}+\left(L_{XX}^{(\omega)}\right)^2d_{31}\right)\cos^2\phi+2\left(L_{YY}^{(\omega)}\right)^2d_{31}\sin^2\phi\right] \end{pmatrix}. \tag*{(10)}

The nonlinear polarization is then projected onto the pp- and ss-polarized detection channels. For the 45∘45^\circ reflection geometry, we find

Pp(2)=PX(2)−PZ(2)2,Ps(2)=PY(2).(11)P_p^{(2)} = \frac{P_X^{(2)}-P_Z^{(2)}}{\sqrt{2}}, \qquad P_s^{(2)} = P_Y^{(2)}. \tag*{(11)}

The detected equilibrium SHG intensity is proportional to the squared magnitude of the corresponding projected nonlinear polarization, Iσ∝∣Pσ(2)∣2I_\sigma\propto\lvert P_\sigma^{(2)}\rvert^2, with σ∈{p,s}\sigma\in\{p,s\}.

(3) process

In addition to the intrinsic second-order SHG response, the MIR electric field can generate a field-dependent nonlinear polarization through the third-order susceptibility. This contribution is linear in the MIR electric field. Since the MIR frequency Ω\Omega is much smaller than the optical probe frequency ω\omega, we treat the MIR field as slowly varying on the optical timescale and describe its effect as a time-dependent modulation of the nonlinear polarization near 2ω2\omega:

Pi(3)(2ω,t)=χijkl(3)Ejprobe(ω)Ekprobe(ω)Elpump(t).(12)P_i^{(3)}(2\omega,t)=\chi_{ijkl}^{(3)}E_j^{\mathrm{probe}}(\omega)E_k^{\mathrm{probe}}(\omega)E_l^{\mathrm{pump}}(t). \tag*{(12)}

For convenience, for each pump-field direction ll we write the corresponding third-order susceptibility as a 3×63 \times6 matrix χl(3)\chi_l^{(3)} acting on the contracted quadratic probe-field vector. The third-order nonlinear polarization can then be written as

P(3)=(E1pumpχ1(3)+E2pumpχ2(3)+E3pumpχ3(3))((E1probe)2(E2probe)2(E3probe)22E2probeE3probe2E1probeE3probe2E1probeE2probe).(13)\mathbf{P}^{(3)}=\left(E_1^{\mathrm{pump}}\chi_1^{(3)}+E_2^{\mathrm{pump}}\chi_2^{(3)}+E_3^{\mathrm{pump}}\chi_3^{(3)}\right) \begin{pmatrix} (E_1^{\mathrm{probe}})^2\\ (E_2^{\mathrm{probe}})^2\\ (E_3^{\mathrm{probe}})^2\\ 2E_2^{\mathrm{probe}}E_3^{\mathrm{probe}}\\ 2E_1^{\mathrm{probe}}E_3^{\mathrm{probe}}\\ 2E_1^{\mathrm{probe}}E_2^{\mathrm{probe}} \end{pmatrix}. \tag*{(13)}

For 4mm4mm symmetry, the corresponding matrices for the three pump-field directions are

χ1(3)=(d11(1)d12(1)d13(1)00000000d26(1)0000d35(1)0),χ2(3)=(00000d26(1)d12(1)d11(1)d13(1)000000d35(1)00),χ3(3)=(0000d15(3)0000d15(3)00d31(3)d31(3)d33(3)000).(14)\begin{aligned} \chi_1^{(3)}&= \begin{pmatrix} d_{11}^{(1)} & d_{12}^{(1)} & d_{13}^{(1)} & 0 & 0 & 0\\ 0 & 0 & 0 & 0 & 0 & d_{26}^{(1)}\\ 0 & 0 & 0 & 0 & d_{35}^{(1)} & 0 \end{pmatrix},\\ \chi_2^{(3)}&= \begin{pmatrix} 0 & 0 & 0 & 0 & 0 & d_{26}^{(1)}\\ d_{12}^{(1)} & d_{11}^{(1)} & d_{13}^{(1)} & 0 & 0 & 0\\ 0 & 0 & 0 & d_{35}^{(1)} & 0 & 0 \end{pmatrix},\\ \chi_3^{(3)}&= \begin{pmatrix} 0 & 0 & 0 & 0 & d_{15}^{(3)} & 0\\ 0 & 0 & 0 & d_{15}^{(3)} & 0 & 0\\ d_{31}^{(3)} & d_{31}^{(3)} & d_{33}^{(3)} & 0 & 0 & 0 \end{pmatrix}. \tag*{(14)} \end{aligned}

The subscripts 1, 2, and 3 on χi(3)\chi^{(3)}_{i} denote the direction of the MIR electric field along the local crystal axes X1X_{1}, X2X_{2}, and X3X_{3}, respectively. In the effective skyrmion model, these axes coincide with the sample axes XX, YY, and ZZ.

For clarity, Fresnel factors are omitted in the following higher-order symmetry analysis. At fixed probe wavelength and geometry, they renormalize the effective coefficients but do not alter the angular functional forms derived below.

Because the MIR pump is normally incident, its electric field lies in the sample plane. We first consider a MIR field polarized along X=X1X=X_{1}, i.e., parallel to the SHG plane of incidence. Using the probe-field components defined above gives

P(3)(ϕ)=Epump(E0probe)2(12(d11(1)+d13(1))cos⁡2ϕ+d12(1)sin⁡2ϕ12d26(1)sin⁡(2ϕ)d35(1)cos⁡2ϕ).(15)\mathbf{P}^{(3)}(\phi)=E^{\mathrm{pump}}\left(E_{0}^{\mathrm{probe}}\right)^{2} \begin{pmatrix} \frac{1}{2}\left(d_{11}^{(1)}+d_{13}^{(1)}\right)\cos^{2}\phi+d_{12}^{(1)}\sin^{2}\phi\\ \frac{1}{\sqrt{2}}d_{26}^{(1)}\sin(2\phi)\\ d_{35}^{(1)}\cos^{2}\phi \end{pmatrix}. \tag*{(15)}

After projection onto the detector channels, the angular dependences can therefore be written as

Pp(3)(ϕ)∝Acos⁡2ϕ+Bsin⁡2ϕ,Ps(3)(ϕ)∝sin⁡(2ϕ),(16)\begin{aligned} P_{p}^{(3)}(\phi)&\propto A\cos^{2}\phi+B\sin^{2}\phi,\\ P_{s}^{(3)}(\phi)&\propto\sin(2\phi), \tag*{(16)} \end{aligned}

where AA and BB are effective coefficients containing the relevant tensor and optical factors.

For a MIR field polarized along Y=X2Y=X_{2}, perpendicular to the SHG plane of incidence, the fourfold symmetry gives

Pp(3)(ϕ)∝sin⁡(2ϕ),Ps(3)(ϕ)∝Ccos⁡2ϕ+Dsin⁡2ϕ,(17)\begin{aligned} P_{p}^{(3)}(\phi)&\propto\sin(2\phi),\\ P_{s}^{(3)}(\phi)&\propto C\cos^{2}\phi+D\sin^{2}\phi, \tag*{(17)} \end{aligned}

where CC and DD are the corresponding effective coefficients.

(4) process

We also consider a higher-order contribution that is quadratic in the MIR electric field and is described by the fourth-order nonlinear susceptibility,

Pi(4)(2ω,t)=χijk,lm(4)Ejprobe(ω)Ekprobe(ω)Elpump(t)Empump(t).(18)P_{i}^{(4)}(2\omega,t)=\chi_{ijk,lm}^{(4)}E_{j}^{\mathrm{probe}}(\omega)E_{k}^{\mathrm{probe}}(\omega)E_{l}^{\mathrm{pump}}(t)E_{m}^{\mathrm{pump}}(t). \tag*{(18)}

For a MIR field polarized along X=X1X=X_{1}, only the l=m=1l=m=1 pump-field component contributes, so the expression reduces to Pi(4)=(Epump)2χijk,11(4)EjprobeEkprobeP_{i}^{(4)}=(E^{\mathrm{pump}})^{2}\chi_{ijk,11}^{(4)}E_{j}^{\mathrm{probe}}E_{k}^{\mathrm{probe}}. For 4mm4mm symmetry, the relevant contracted tensor can be written as

(χijk,11(4))=(0000d15(11)0000d24(11)00d31(11)d32(11)d33(11)000).(19)\left(\chi^{(4)}_{ijk,11}\right)= \begin{pmatrix} 0 & 0 & 0 & 0 & d^{(11)}_{15} & 0 \\ 0 & 0 & 0 & d^{(11)}_{24} & 0 & 0 \\ d^{(11)}_{31} & d^{(11)}_{32} & d^{(11)}_{33} & 0 & 0 & 0 \end{pmatrix}. \tag*{(19)}

Using the same probe geometry gives

P(4)(ϕ)=(Epump)2(E0probe)2(d15(11)cos⁡2ϕ12d24(11)sin⁡(2ϕ)12(d31(11)+d33(11))cos⁡2ϕ+d32(11)sin⁡2ϕ).(20)\mathbf{P}^{(4)}(\phi)=\left(E^{\mathrm{pump}}\right)^2\left(E_0^{\mathrm{probe}}\right)^2 \begin{pmatrix} d^{(11)}_{15}\cos^2\phi\\ \frac{1}{\sqrt{2}}d^{(11)}_{24}\sin(2\phi) \\ \frac{1}{2}\left(d^{(11)}_{31}+d^{(11)}_{33}\right)\cos^2\phi+d^{(11)}_{32}\sin^2\phi \end{pmatrix}. \tag*{(20)}

The corresponding detector-channel dependences are therefore

Pp(4)(ϕ)∝A′cos⁡2ϕ+B′sin⁡2ϕ,P_p^{(4)}(\phi)\propto A'\cos^2\phi+B'\sin^2\phi,
Ps(4)(ϕ)∝sin⁡(2ϕ),(21)P_s^{(4)}(\phi)\propto\sin(2\phi), \tag*{(21)}

where A′A' and B′B' are effective coefficients.

For a MIR field polarized along Y=X2Y=X_2, the corresponding response follows from the 90° rotational symmetry of the 4mm4mm crystal. The tensor coefficients are transformed accordingly, while the detector-channel angular dependences retain the same functional form,

Pp(4)(ϕ)∝C′cos⁡2ϕ+D′sin⁡2ϕ,P_p^{(4)}(\phi)\propto C'\cos^2\phi+D'\sin^2\phi,
Ps(4)(ϕ)∝sin⁡(2ϕ),(22)P_s^{(4)}(\phi)\propto\sin(2\phi), \tag*{(22)}

with C′C' and D′D' denoting the corresponding effective coefficients.

Pump polarization dependence

Polar texture dependence

All polar phases show a strong light-induced SHG, as plotted in Figure S6. The transient SHG intensity from the polar skyrmion phase is about 3 times stronger than those from the a1_1/a2_2 and the mixed phase samples.

Absence of THz emission

We have conducted electro-optical sampling measurement to characterize the wavelength of the MIR pump and to investigate potential pump-induced optical rectification. The results are shown in Figure S7. A weak THz signal is seen from the polar skyrmion (Figure S7, Figure S7). However, This THz signal is identical to the one measured with a gold mirror (Figure S7, Figure S7) instead of the polar skyrmion. In addition, when placing a MIR filter in front or behind the sample, the THz signal shows no difference (Figure S7). These observations indicate that the THz signal is a residual from the mid-infrared generation. There is no far-field THz emission generated by the sample.

Based on the simplified 4mm4mm model (see the χ(2)\chi^{(2)} tensor as shown in Eq. (7)), we obtain the polarization induced by optical rectification of the pump field as:

P(0)∣EMIR∥X1=(0000d150000d1500d31d31d33000)(EMIR200000)=(00d31EMIR2)(23)\left.\mathbf{P}(0)\right|_{E_{\mathrm{MIR}}\parallel X_1} = \begin{pmatrix} 0 & 0 & 0 & 0 & d_{15} & 0 \\ 0 & 0 & 0 & d_{15} & 0 & 0 \\ d_{31} & d_{31} & d_{33} & 0 & 0 & 0 \end{pmatrix} \begin{pmatrix} E_{\mathrm{MIR}}^2 \\ 0 \\ 0 \\ 0 \\ 0 \\ 0 \end{pmatrix} = \begin{pmatrix} 0 \\ 0 \\ d_{31}E_{\mathrm{MIR}}^2 \end{pmatrix} \tag*{(23)}
P(0)∣EMIR∥X2=(0000d150000d1500d31d31d33000)(0EMIR20000)=(00d31EMIR2)(24)\left.\mathbf{P}(0)\right|_{E_{\mathrm{MIR}}\parallel X_2} = \begin{pmatrix} 0 & 0 & 0 & 0 & d_{15} & 0 \\ 0 & 0 & 0 & d_{15} & 0 & 0 \\ d_{31} & d_{31} & d_{33} & 0 & 0 & 0 \end{pmatrix} \begin{pmatrix} 0 \\ E_{\mathrm{MIR}}^2 \\ 0 \\ 0 \\ 0 \\ 0 \end{pmatrix} = \begin{pmatrix} 0 \\ 0 \\ d_{31}E_{\mathrm{MIR}}^2 \end{pmatrix} \tag*{(24)}

Both cases give the same result on optical rectification induced polarization.

Heterodyned and homodyned SHG components

For processes which involve the third-order nonlinearity, the pump electric field E(ω1)E(\omega_1) brings in an additional contribution to the second-harmonic polarization P(2ω±ω1)∝χ(3)E(ω1)E(ω)E(ω)P(2\omega\pm\omega_1) \propto\chi^{(3)}E(\omega_1)E(\omega)E(\omega). If the material has a nonzero χ(2)\chi^{(2)}, the SHG intensity is given by

I∝{[χ(2)+χ(3)E(ω1)]E(ω)E(ω)}2(25)I \propto\left\{\left[\chi^{(2)}+\chi^{(3)}E(\omega_1)\right]E(\omega)E(\omega)\right\}^{2} \tag*{(25)}

This equation has three terms: the equilibrium SHG intensity, a heterodyned term oscillates at the pump frequency ω1\omega_1 (observed in non-centrosymmetric materials[44, 54]), and a homodyned term containing [χ(3)E(ω1)]2\left[\chi^{(3)}E(\omega_1)\right]^2 which includes a displacive response at 2ω2\omega and a 2ω12\omega_1 oscillatory term (observed in centrosymmetric systems with zero χ(2)\chi^{(2)}[44]). All these terms are summarized in Table S1.

TermDescriptionFrequencyPump field dependence
[χ(2)E(ω)E(ω)]2≡E02\left[\chi^{(2)}E(\omega)E(\omega)\right]^2 \equiv E_0^2Equilibrium2ω2\omegaN/A
2χ(2)χ(3)E(ω1)[E(ω)E(ω)]2≡2E0E(χ(3))2\chi^{(2)}\chi^{(3)}E(\omega_1)\left[E(\omega)E(\omega)\right]^2 \equiv2E_0E(\chi^{(3)})Heterodyne2ω±ω12\omega\pm\omega_11E(ω1)E(\omega_1)
[χ(3)E(ω1)E(ω)E(ω)]2≡E2(χ(3))\left[\chi^{(3)}E(\omega_1)E(\omega)E(\omega)\right]^2 \equiv E^2(\chi^{(3)})Homodyne2ω,2ω±2ω12\omega, 2\omega\pm2\omega_12E2(ω1)E^2(\omega_1)

Table S1. Heterodyned and homodyned SHG components involving χ(2)\chi^{(2)} and χ(3)\chi^{(3)}.

Following the notation in Table S1, we now discuss the processes involving the fourth-order optical susceptibility. In this case, the second-harmonic generation intensity is given by

I∝{[χ(2)+χ(3)E(ω1)+χ(4)E(ω1)E(ω1)]E(ω)E(ω)}2(26)I \propto\left\{\left[\chi^{(2)}+\chi^{(3)}E(\omega_1)+\chi^{(4)}E(\omega_1)E(\omega_1)\right]E(\omega)E(\omega)\right\}^{2} \tag*{(26)}

All terms contributing to the SHG intensity are listed in Table S2. Oscillatory components at frequencies higher than the pump frequency (ω1\omega_{1}) are removed by the bandpass filter in front of the photomultiplier. This table reveals that the quasi-static component of the signal depends on E2(ω1)E^{2}(\omega_{1}) and E4(ω1)E^{4}(\omega_{1}), and that the amplitude of the oscillation at ω1\omega_{1} exhibits dependence on E(ω1)E(\omega_{1}) and E3(ω1)E^{3}(\omega_{1}).

TermDescriptionFrequencyPump field dependence
E02E_{0}^{2}Equilibrium2ω2\omegaN/A
2E0E(χ(3))2E_{0}E(\chi^{(3)})Heterodyne2ω±ω12\omega\pm\omega_{1}E(ω1)E(\omega_{1})
2E0E(χ(4))2E_{0}E(\chi^{(4)})Heterodyne2ω,2ω±2ω12\omega, 2\omega\pm2\omega_{1}E2(ω1)E^{2}(\omega_{1})
2E(χ(3))E(χ(4))2E(\chi^{(3)})E(\chi^{(4)})Heterodyne2ω±ω1,2ω±3ω12\omega\pm\omega_{1}, 2\omega\pm3\omega_{1}E3(ω1)E^{3}(\omega_{1})
E2(χ(3))E^{2}(\chi^{(3)})Homodyne2ω,2ω±2ω12\omega, 2\omega\pm2\omega_{1}E2(ω1)E^{2}(\omega_{1})
E2(χ(4))E^{2}(\chi^{(4)})Homodyne2ω,2ω±2ω1,2ω±4ω12\omega, 2\omega\pm2\omega_{1}, 2\omega\pm4\omega_{1}E4(ω1)E^{4}(\omega_{1})

Table S2. Heterodyned and homodyned SHG components involving χ(2)\chi^{(2)}, χ(3)\chi^{(3)} and χ(4)\chi^{(4)}. The 2ω2\omega and 2ω±ω12\omega\pm\omega_{1} terms are measured, while higher frequency sidebands are removed by the bandpass filter.

Alternatively, the third term in Eq. (26), χ(4)E(ω1)E(ω1)E(ω)E(ω)\chi^{(4)}E(\omega_{1})E(\omega_{1})E(\omega)E(\omega), can be replaced by a cascaded χ(2)\chi^{(2)} and χ(3)\chi^{(3)} process, χ(3)E(ω)E(ω)P(0)\chi^{(3)}E(\omega)E(\omega)P(0) with P(0)=χ(2)E(ω1)E(−ω1)P(0) = \chi^{(2)}E(\omega_{1})E(-\omega_{1}). Therefore, the χ(4)\chi^{(4)} term in Table S2 corresponds to either an intrinsic or an effective χ(4)\chi^{(4)} process.

For an effective χ(4)\chi^{(4)} generated via cascaded optical rectification and a subsequent χ(3)\chi^{(3)} process, the rectification-induced polarization field remains perpendicular to the sample plane upon a 90° pump polarization rotation (see Eqs. (23),(24)). Thus, the E2(χ(4))E^{2}(\chi^{(4)}) term in Table S2 remains independent of the pump polarization. In contrast, the E2(χ(3))E^{2}(\chi^{(3)}) term varies with the pump polarization, which explains the experimental observation (Figure S5).

Pump polarization and sample orientation dependence

Figure S5. Pump polarization and sample orientation dependence. Transient SHG polarimetry probed in two orthogonal channels at t=0t = 0 ps with three different geometries: a, the initial geometry defined in main text Figure 1a and Figure S2. b, sample rotated by 45∘45^\circ along the surface normal, and c, MIR polarization rotated 90∘90^\circ. The SHG intensity in all cases are normalized by the maximum transient intensity ΔIp\Delta I_p in case a.

In the experiment, we measured the change in second-harmonic generation (SHG) intensity as a function of the pump–probe time delay. A representative pump–probe trace of the polar skyrmion sample is shown in Figure S8. The transient SHG intensity exhibits a pronounced enhancement accompanied by oscillations at the pump frequency. The displacive response is defined as the non-oscillatory component of the transient signal, which is modeled using a Gaussian function

Plot showing Normalized $\Delta I_p$, Displacive response, and Oscillatory response; axes labeled Normalized $\Delta I_p(2\omega,\phi=4^\circ)$ and Time delay (ps)

Figure S8. Extraction of displacive and oscillatory responses from experimental data. With the MIR pumping, the transient SHG intensity of the polar skyrmion sample shows a giant enhancement with oscillations at the pump frequency. The non-oscillatory contribution is modeled using a Gaussian function and identified as the displacive response. The oscillatory response is obtained by subtracting this fitted displacive component from the raw data.

y(x)=Aexp⁡[−(x−x02σ)2](27)y(x) = A \exp\left[-\left(\frac{x-x_{0}}{\sqrt{2}\sigma}\right)^{2}\right] \tag*{(27)}

where AA is the peak amplitude, x0x_{0} is the peak timing, and σ\sigma is the standard deviation of the Gaussian distribution. The time traces were acquired over multiple scans. Each scan was independently fitted with a Gaussian function, and the standard deviation of the extracted peak displacive response across all scans was taken as the uncertainty for a given pump fluence. The oscillatory component was obtained by subtracting the averaged displacive response from the averaged raw time-trace at the same pump fluence. We then evaluated the Fourier transform amplitude of the oscillatory component. The oscillation amplitude was extracted from the area of the corresponding frequency-domain peak using a Gaussian function (Eq. (27)) in the frequency domain, where x0x_{0} is the peak center frequency. The peak area is Aσ2πA\sigma\sqrt{2\pi}. The uncertainty in AA is estimated via standard error propagation, including contributions from spectral background noise, Fourier-transform noise, and the amplitude fitting uncertainty derived from the corresponding confidence interval. The uncertainty in σ\sigma is estimated from the 90% confidence interval of the fit.

Table S2 shows that the χ(3)\chi^{(3)} contributes to a transient SHG intensity oscillation at the pump frequency that is linearly proportional to the pump field. In addition, χ(3)\chi^{(3)} and χ(4)\chi^{(4)} contribute to a cubic pump-field dependence of the oscillation at the pump frequency. Therefore, analyzing the pump-field dependence of the SHG intensity oscillation provides access to higher-order optical susceptibilities. Figure S9 presents the same data as Figure 4 in the main text and highlights the individual contributions of the linear and cubic terms to the oscillatory response. The χ(4)\chi^{(4)} contribution becomes prominent at high pump fluences. Table S3 shows the fitting parameters.

In addition, we have tested whether the pump-field dependence for three polar textures is better described by a single global mixed order scaling model (aEMIR+bEMIR3)(aE_{\mathrm{MIR}} + bE_{\mathrm{MIR}}^{3}) over the entire measured field range, or a two-regime (piecewise) description is preferred. We used the small-sample correct Akaike information criterion (AICc) as the primary model-selection metric. In addition to model selection, we computed diagnostics from the global mixed fits to summarize the relative importance of the low- and high-order terms and the stability of inferred crossover scales. The conclusions are:

  1. The oscillatory responses for the polar skyrmion phase and the a1/a2a_{1}/a_{2} phase are both well described by a single global mixed odd-order model (aEMIR+bEMIR3)(aE_{\mathrm{MIR}}+bE_{\mathrm{MIR}}^{3}) over the measured field range. In both cases, piecewise alternatives are not preferred after AICc penalization, and this conclusion remains stable across the tested weighting scan.

  1. The oscillatory response of the mixed a1/a2a_{1}/a_{2} and polar vortex phase is best described by a simpler single-EE global model. The cubic contribution remains weak over the measured range, the mixed-model crossover is pushed beyond the measured field range, and piecewise alternatives are not competitive.

Pump wavelength dependence

The peak response of the transient SHG intensity versus pump photon energy for the polar skyrmion is shown in Figure S10. According to Ref. [46], the highest A1_1 phonon in PbTiO3_3 is located at 646 cm−1^{-1} (or 19.4 THz, TO frequency), for which our MIR pump could potentially resonantly excite. However, the transient SHG does not peak at this frequency, but is close to the LO phonon frequency[46].

Figure S10. Pump wavelength dependence. The normalized displacive response in ΔIp(2ω)\Delta I_p(2\omega) as a function of the excitation photon energy. Vertical error bars represent the standard deviation. Horizontal bars are the bandwidths of the mid-infrared pump pulses.

Phase-field simulations

The phase-field formulation used to model the PTO/STO superlattice defines the free energy density as follows:

f=αiPi2+αijPi2Pj2+αijkPi2Pj2Pk2+12cijklεijεkl−qijklεijPkPl+12GijklPi,jPk,l−12KijEiEj−EiPi.(28)\begin{aligned} f ={}& \alpha_i P_i^2 + \alpha_{ij}P_i^2P_j^2 + \alpha_{ijk}P_i^2P_j^2P_k^2 + \frac{1}{2}c_{ijkl}\varepsilon_{ij}\varepsilon_{kl} - q_{ijkl}\varepsilon_{ij}P_kP_l \\ &+ \frac{1}{2}G_{ijkl}P_{i,j}P_{k,l} - \frac{1}{2}K_{ij}E_iE_j - E_iP_i. \tag*{(28)} \end{aligned}

In this expression, αi\alpha_i, αij\alpha_{ij}, and αijk\alpha_{ijk} represent the Landau coefficients accounting for the thermodynamic expansion of the free energy; cijklc_{ijkl} are the elastic stiffness parameters; qijkl=cijmnQmnklq_{ijkl}=c_{ijmn}Q_{mnkl}, where QijklQ_{ijkl} is the electrostrictive tensor, capture the electrostrictive coupling between strain and polarization; GijklG_{ijkl} are associated with the polarization gradient energy; and Kij=κδijK_{ij}=\kappa\delta_{ij} denote the background dielectric permittivity tensor components. The variables PiP_i, εij\varepsilon_{ij}, EiE_i, and uiu_i correspond to the components of polarization, mechanical strain, electric field, and displacement, respectively.

The elastic strain is given by εijelas=εij−εij0\varepsilon_{ij}^{\mathrm{elas}}=\varepsilon_{ij}-\varepsilon_{ij}^{0}, where εij\varepsilon_{ij} is the total strain and εij0=QijklPkPl\varepsilon_{ij}^{0}=Q_{ijkl}P_kP_l is the eigen strain. The total strain is composed of two parts: εij=εijext+ηij\varepsilon_{ij}=\varepsilon_{ij}^{\mathrm{ext}}+\eta_{ij}, where εijext\varepsilon_{ij}^{\mathrm{ext}} is the homogeneous strain (the external mechanical strain), and ηij=12(ui,j+uj,i)\eta_{ij}=\frac{1}{2}(u_{i,j}+u_{j,i}) is the inhomogeneous strain. The total electric field EiE_i includes the effective THz field from our pump and the internal fields. Summation is implied over repeated indices, following the Einstein convention across spatial dimensions xix_i (i=1,2,3i=1,2,3), and partial derivatives with respect to spatial coordinates are denoted by subscript “, ii”.

The temporal evolution of the polarization field is governed by the polarization dynamic equation:[51, 52]

μ∂2Pi∂t2+γ∂Pi∂t=−δFδPi(29)\mu\frac{\partial^2P_i}{\partial t^2}+\gamma\frac{\partial P_i}{\partial t}=-\frac{\delta F}{\delta P_i} \tag*{(29)}

Here, μ\mu is the effective inertial coefficient of polarization, γ\gamma is the damping constant, and F=∫f dVF=\int f\,dV represents the total free energy over the simulated volume[27]. In addition, the system satisfies both the mechanical equilibrium condition:

σij,j=∂∂xj(∂f∂εij)=0(30)\sigma_{ij,j} = \frac{\partial}{\partial x_j}\left(\frac{\partial f}{\partial\varepsilon_{ij}}\right) = 0 \tag*{(30)}

and the electrostatic Maxwell’s equation:

Di,i=−∂∂xi(∂f∂Ei)=0.(31)D_{i,i} = -\frac{\partial}{\partial x_i}\left(\frac{\partial f}{\partial E_i}\right) = 0. \tag*{(31)}

These conditions hold under the assumptions of a body-force-free and charge-free ferroelectric medium, where σij\sigma_{ij} and DiD_i are the stress and electric displacement components, respectively. To numerically solve the resulting coupled partial differential equations, a semi-implicit Fourier-spectral method [55] is employed, offering high computational efficiency and accuracy in handling spatial derivatives.

The material parameters used for PTO and STO in the simulations are taken from references [27, 56, 57, 58], from which we take an effective polarization mass density μ\mu of 6.27×10−186.27 \times10^{-18} kg·m3^3/C2^2 and a damping coefficient γ\gamma of 2×10−72 \times10^{-7} Ω⋅\Omega\cdot m. This effective polarization mass density reproduces the frequencies of the PbTiO3_3 E(TO) and A1_1(TO) modes. To initiate the simulation, a small random fluctuation in polarization—less than 0.01P00.01P_0, where P0=0.76P_0 = 0.76 C/m2^2 is the spontaneous polarization of PTO at room temperature—is introduced as the initial condition. The simulations were done for room temperature (293 K).

To simulated three polar phases, we used the same coefficients listed in Table S4 for the free energy density. The only parameter varied was the applied biaxial strain. For the a1_1/a2_2 and skyrmion phases, we assumed that STO is strain-free, while PTO experiences a biaxial strain of εxx=εyy=0%\varepsilon_{xx} = \varepsilon_{yy} = 0\% for the a1_1/a2_2 phase and εxx=εyy=−1.34%\varepsilon_{xx} = \varepsilon_{yy} = -1.34\% for the skyrmion phase. In the case of the vortex phase, the strain is anisotropic. For STO, the strain components are εxx=1.2%\varepsilon_{xx} = 1.2\% and εyy=1.05%\varepsilon_{yy} = 1.05\%, while for PTO they are εxx=−0.12%\varepsilon_{xx} = -0.12\% and εyy=−0.275%\varepsilon_{yy} = -0.275\%. We simulated the pure vortex phase for simplicity. We evaluated the in-plane permittivity for two cases, with an in-plane electric field applied perpendicular to (Figure S11) or parallel to the vortex core (Figure S11). For both cases, the local permittivity is predominantly positive, in contrast to the large alternating positive and negative local permittivity in the polar skyrmion case. As the domain boundaries are where large local permittivity arises, we anticipate the a1_1/a2_2 and vortex mixed phase to have a larger effective permittivity than either the pure vortex and the pure a1_1/a2_2 phase.

CoefficientPTOSTO
Landau coefficient (Temperature TT in K)
α1\alpha_1 (10710^7 C−2^{-2}m2^2N)0.038(T−752)0.038(T - 752)4.05[coth⁡(54/T)−1.056]4.05[\coth(54/T) - 1.056]
α11\alpha_{11} (10810^8 C−4^{-4}m6^6N)−0.73-0.731.71.7
α12\alpha_{12} (10910^9 C−4^{-4}m6^6N)0.750.754.454.45
α111\alpha_{111} (10810^8 C−4^{-4}m10^{10}N)2.62.6
α112\alpha_{112} (10810^8 C−4^{-4}m10^{10}N)6.16.1
α123\alpha_{123} (10910^9 C−4^{-4}m10^{10}N)−3.7-3.7
Gradient energy coefficient
G11G_{11} (10−1010^{-10} C−2^{-2}m4^4N)1.0381.0381.0381.038
G12G_{12} (10−1010^{-10} C−2^{-2}m4^4N)0000
G44G_{44} (10−1010^{-10} C−2^{-2}m4^4N)0.5190.5190.5190.519
Elastic stiffness
c11c_{11} (10910^9 Pa)230230330330
c12c_{12} (10910^9 Pa)100100100100
c44c_{44} (10910^9 Pa)7070125125
Electrostrictive coupling coefficient
q11q_{11} (101010^{10} C−2^{-2}m2^2N)1.1421.1421.261.26
q12q_{12} (10910^9 C−2^{-2}m2^2N)0.490.49−13-13
q44q_{44} (10910^9 C−2^{-2}m2^2N)3.7183.71811.9611.96
Q11Q_{11} (C−2^{-2}m4^4)0.0890.0890.04570.0457
Q12Q_{12} (C−2^{-2}m4^4)−0.026-0.026−0.0135-0.0135
Q44Q_{44} (C−2^{-2}m4^4)0.06750.06750.009570.00957
Background dielectric constant (ϵ0=8.85×10−12\epsilon_0 = 8.85 \times10^{-12} Fm−1^{-1})
κb\kappa^b (ϵ0\epsilon_0)10101010

Table S4. Parameters used for simulating three phases.

Phase-field simulations of the local permittivity distribution for the polar vortex phase

Figure S11. Phase-field simulations of the local permittivity distribution for the polar vortex phase. Ferroelectric polarization distribution (arrows) and the local permittivity (color coded) of the PTO layer for the pure vortex phase. a, The in-plane permittivity, ϵxx\epsilon_{xx}. b), The in-plane permittivity, ϵyy\epsilon_{yy}.

References

  1. [1]Shen, Y.R.: The Principles of Nonlinear Optics. Wiley-Interscience, New Jersey (2002)
  2. [2]Boyd, R.W.: Nonlinear Optics. Academic Press, New York (2008)
  3. [3]Fiebig, M., Pavlov, V.V., Pisarev, R.V.: Second-harmonic generation as a tool for studying electronic and magnetic structures of crystals: review. J. Opt. Soc. Am. B 22, 96–117 (2005). https://doi.org/10.1364/JOSAB.22.000096
  4. [4]Denev, S.A., Lummen, T.T.A., Barnes, E., Kumar, A., Gopalan, V.: Probing ferroelectrics using optical second harmonic generation. J. Am. Ceram. Soc. 94, 2699–2727 (2011). https://doi.org/10.1111/j.1551-2916.2011.04740.x
  5. [5]Kauranen, M., Zayats, A.V.: Nonlinear plasmonics. Nat. Photonics 6, 737 (2012). https://doi.org/10.1038/NPHOTON.2012.244
  6. [6]Pettine, J., Padmanabhan, P., Shi, T., Gingras, L., McClintock, L., Chang, C.-C., Kwock, K.W.C., Yuan, L., Huang, Y., Nogan, J., Baldwin, J.K., Adel, P., Holzwarth, R., Azad, A.K., Ronning, F., Taylor, A.J., Prasankumar, R.P., Lin, S.-Z., Chen, H.-T.: Light-driven nanoscale vectorial currents. Nature 626, 984–989 (2024). https://doi.org/10.1038/s41586-024-07037-4
  7. [7]Franken, P.A., Hill, A.E., Peters, C.W., Weinreich, G.: Generation of optical harmonics. Phys. Rev. Lett. 7, 118–119 (1961). https://doi.org/10.1103/PhysRevLett.7.118
  8. [8]Guyot-Sionnest, P., Hunt, J.H., Shen, Y.R.: Sum-frequency vibrational spectroscopy of a langmuir film: Study of molecular orientation of a two-dimensional system. Phys. Rev. Lett. 59, 1597–1600 (1987). https://doi.org/10.1103/PhysRevLett.59.1597
  9. [9]Manzoni, C., Cerullo, G.: Design criteria for ultrafast optical parametric amplifiers. J. Opt. 18, 103501 (2016). https://doi.org/10.1088/2040-8978/18/10/103501
  10. [10]New, G.H.C., Ward, J.F.: Optical third-harmonic generation in gases. Phys. Rev. Lett. 19, 556–559 (1967). https://doi.org/10.1103/PhysRevLett.19.556
  11. [11]Hendry, E., Hale, P.J., Moger, J., Savchenko, A.K., Mikhailov, S.A.: Coherent nonlinear optical response of graphene. Phys. Rev. Lett. 105, 097401 (2010). https://doi.org/10.1103/PhysRevLett.105.097401
  12. [12]Yue, L., Liu, C., Han, S., Hong, H., Wang, Y., Liu, Q., Qi, J., Li, Y., Wu, D., Liu, K., Wang, E., Dong, T., Wang, N.: Giant nonlinear optical wave mixing in a van der waals correlated insulator. Sci. Adv. 10, 6216 (2024). https://doi.org/10.1126/sciadv.adn6216
  13. [13]Shkurinov, A.P., Dubrov­skii, A.V., Koroteev, N.I.: Second harmonic generation in an optically active liquid: Experimental observation of a fourth-order optical nonlinearity due to molecular chirality. Phys. Rev. Lett. 70, 1085–1088 (1993). https://doi.org/10.1103/PhysRevLett.70.1085
  14. [14]Hafez, H.A., Kovalev, S., Deinert, J.-C., Mics, Z., Green, B., Awari, N., Chen, M., Germansky, S., Lehnert, U., Teichert, J., Wang, Z., Tielrooij, K.-J., Liu, Z., Chen, Z., Narita, A., Müllen, K., Bonn, M., Gensch, M., Turchinovich, D.: Extremely efficient terahertz high-harmonic generation in graphene by hot dirac fermions. Nature 561, 507–511 (2018). https://doi.org/10.1038/s41586-018-0508-1
  15. [15]Dutt, A., Mohanty, A., Gaeta, A.L., Lipson, M.: Nonlinear and quantum photonics using integrated optical materials. Nat. Rev. Mater. 9, 321–346 (2024). https://doi.org/10.1038/s41578-024-00668-z
  16. [16]Liberal, I., Engheta, N.: Near-zero refractive index photonics. Nat. Photonics 11, 149–158 (2017). https://doi.org/10.1038/nphoton.2017.13
  17. [17]Sun, H., Chen, P., Mao, W., Guo, C., Li, Y., Wang, J., Sun, W., Xu, D., Hao, B., Zhang, T., Ma, J., Yang, J., Cao, Z., Yan, S., Guan, Y., Wen, Z., Mao, Z., Zheng, N., Gu, Z., Huang, H., Wang, P., Zhang, Y., Wu, D., Nie, Y.: Ferroelectric topologies in BaTiO₃ nanomembranes for light field manipulation. Nat. Nanotechnol 534, 524–528 (2025). https://doi.org/10.1038/s41565-025-01919-y
  18. [18]Junquera, J., Nahas, Y., Prokhorenko, S., Bellaiche, L., Íñiguez, J., Schlom, D.G., Chen, L.-Q., Salahuddin, S., Muller, D.A., Martin, L.W., Ramesh, R.: Topological phases in polar oxide nanostructures. Rev. Mod. Phys. 95, 025001 (2023). https://doi.org/10.1103/RevModPhys.95.025001
  19. [19]Das, S., Tang, Y.L., Hong, Z., Gonçalves, M.A.P., McCarter, M.R., Klewe, C., Nguyen, K.X., Gómez-Ortiz, F., Shafer, P., Arenholz, E., Stoica, V.A., Hsu, S.-L., Wang, B., Ophus, C., Liu, J.F., Nelson, C.T., Prasad, S.S.B., Mei, A.B., Schlom, D.G., Íñiguez, J., García-Fernández, P., Muller, D.A., Chen, L.Q., Junquera, J., Martin, L.W., Ramesh, R.: Observation of room-temperature polar skyrmions. Nature 568, 368–372 (2019). https://doi.org/10.1038/s41586-019-1092-8
  20. [20]Wang, Y.J., Feng, Y.P., Zhu, Y.L., Tang, Y.L., Yang, L.X., Zou, M.J., Geng, W.R., Han, M.J., Guo, X.W., Wu, B., Ma, X.L.: Polar meron lattice in strained oxide ferroelectrics. Nature Mater. 19, 881–886 (2020). https://doi.org/10.1038/s41563-020-0694-8
  21. [21]Yadav, A.K., Nelson, C.T., Hsu, S.L., Hong, Z., Clarkson, J.D., Schleputz, C.M., Damodaran, A.R., Shafer, P., Arenholz, E., Dedon, L.R., Chen, D., Vishwanath, A., Minor, A.M., Chen, L.Q., Scott, J.F., Martin, L.W., Ramesh, R.: Observation of polar vortices in oxide superlattices. Nature 530, 198–201 (2016). https://doi.org/10.1038/nature16463
  22. [22]Zubko, P., Wojdel, J.C., Hadjimichael, M., Fernandez-Pena, S., Sené, A., Luk’yanchuk, I., Triscone, J.-M., Íñiguez, J.: Negative capacitance in multidomain ferroelectric superlattices. Nature 534, 524–528 (2016). https://doi.org/10.1038/nature17659
  23. [23]Yadav, A.K., Nguyen, K.X., Hong, Z., García-Fernández, P., Aguado-Puente5, P., Nelson, C.T., Das, S., Prasad, B., Kwon, D., Cheema, S., Khan, A.I., Hu, C., Íñiguez, J., Junquera, J., Chen, L.-Q., Muller, D.A., Ramesh, R., Salahuddin, S.: Spatially resolved steady-state negative capacitance. Nature 565, 468–471 (2019). https://doi.org/10.1038/s41586-018-0855-y
  24. [24]Das, S., Hong, Z., Stoica, V.A., Gonçalves, M.A.P., Shao, Y.T., Parsonnet, E., Marksz, E.J., Saremi, S., McCarter, M.R., Reynoso, A., Long, C.J., Hagerstrom, A.M., Meyers, D., Ravi, V., Prasad, B., Zhou, H., Zhang, Z., Wen, H., Gómez-Ortiz, F., García-Fernández, P., Bokor, J., Íñiguez, J., Freeland, J.W., Orloff, N.D., Junquera, J., Chen, L.Q., Salahuddin, S., Muller, D.A., Martin, L.W., Ramesh, R.: Local negative permittivity and topological phase transition in polar skyrmions. Nature Mater. 20, 194–201 (2021). https://doi.org/10.1038/s41563-020-00818-y
  25. [25]Stoica, V.A., Laanait, N., Dai, C., Hong, Z., Yuan, Y., Zhang, Z., Lei, S., McCarter, M.R., Yadav, A., Damodaran, A.R., Das, S., Stone, G.A., Karapetrova, J., Walko, D.A., Zhang, X., Martin, L.W., Ramesh, R., Chen, L.-Q., Wen, H., Gopalan, V., Freeland, J.W.: Optical creation of a supercrystal with three-dimensional nanoscale periodicity. Nature Mater. 18, 377–383 (2019). https://doi.org/10.1038/s41563-019-0311-x
  26. [26]Gao, L., Prokhorenko, S., Nahas, Y., Bellaiche, L.: Dynamical control of topology in polar skyrmions via twisted light. Phys. Rev. Lett. 132, 026902 (2024). https://doi.org/10.1103/PhysRevLett.132.026902
  27. [27]Li, Q., Stoica, V.A., Paściak, M., Zhu, Y., Yuan, Y., Yang, T., McCarter, M.R., Das, S., Yadav, A.K., Park, S., Dai, C., Lee, H.J., Ahn, Y., Marks, S.D., Yu, S., Kadlec, C., Sato, T., Hoffmann, M.C., Chollet, M., Kozina, M.E., Nelson, S., Zhu, D., Walko, D.A., Lindenberg, A.M., Evans, P.G., Chen, L.-Q., Ramesh, R., Martin, L.W., Gopalan, V., Freeland, J.W., Hlinka, J., Wen, H.: Subterahertz collective dynamics of polar vortices. Nature 592, 376–380 (2021). https://doi.org/10.1038/s41586-021-03342-4
  28. [28]Wang, H.H., Stoica, V., Dai, C., Paściak, M., Das, S., Yang, T., Goncalves, M.A.P., Kulda, J., McCarter, M.R., Mangu, A., Cao, Y., Padma, H., Saha, U., Zhu, D., Sato, T., Song, S., Hoffmann, M., Kramer, P., Nelson, S., Sun, Y., Nguyen, Q., Zhang, Z., Ramesh, R., Martin, L., Lindenberg, A.M., Chen, L.-Q., Freeland, J.W., Hlinka, J., Gopalan, V., Wen, H.: Ultrafast excitation of polar skyrmions. Nat. Commun. 16, 8994 (2025). https://doi.org/10.1038/s41467-025-64033-6
  29. [29]Hong, Z., Chen, L.-Q.: Blowing polar skyrmion bubbles in oxide superlattices. Acta Mater. 152, 155–161 (2018). https://doi.org/10.1016/j.actamat.2018.04.022
  30. [30]Nahas, Y., Prokhorenko, S., Zhang, Q., Govinden, V., Valanoor, N., Bellaiche, L.: Topology and control of self-assembled domain patterns in low-dimensional ferroelectrics. Nature Commun. 11, 5779 (2020). https://doi.org/10.1038/s41467-020-19519-w
  31. [31]Han, L., Addiego, C., Prokhorenko, S., Wang, M., Fu, H., Nahas, Y., Yan, X., Cai, S., Wei, T., Fang, Y., Liu, H., Ji, D., Guo, W., Gu, Z., Yang, Y., Wang, P., Bellaiche, L., Chen, Y., Wu, D., Nie, Y., Pan, X.: High-density switchable skyrmion-like polar nanodomains integrated on silicon. Nature 603, 63–67 (2022). https://doi.org/10.1038/s41586-021-04338-w
  32. [32]Zhu, R., Jiang, Z., Zhang, X., Zhong, X., Tan, C., Liu, M., Sun, Y., Li, X., Qi, R., Qu, K., Liu, Z., Wu, M., Li, M., Huang, B., Xu, Z., Wang, J., Liu, K., Gao, P., Wang, J., Bai, J.L.X.: Dynamics of polar skyrmion bubbles under electric fields. Phys. Rev. Lett. 129, 107601 (2022). https://doi.org/10.1103/PhysRevLett.129.107601
  33. [33]Behera, P., May, M.A., Gómez-Ortiz, F., Susarla, S., Das, S., Nelson, C.T., Caretta, L., Hsu, S.-L., McCarter, M.R., Savitzky, B.H., Barnard, E.S., Raja, A., Hong, Z., García-Fernandez, P., Lovesey, S.W., van der Laan, G., Ercius, P., Ophus, C., Martin, L.W., Junquera, J., Raschke, M.B., Ramesh, R.: Electric field control of chirality. Sci. Adv. 8, 8030 (2022). https://doi.org/10.1126/sciadv.abj8030
  34. [34]Wang, S., Li, W., Deng, C., Hong, Z., Gao, H.-B., Li, X., Gu, Y., Zheng, Q., Wu, Y., Evans, P.G., Li, J.-F., Nan, C.-W., Li, Q.: Giant electric field-induced second harmonic generation in polar skyrmions. Nature Commun. 15, 1374 (2024). https://doi.org/10.1038/s41467-024-45755-5
  35. [35]Dogadov, O., Trovatello, C., Yao, B., Soavi, G., Cerullo, G.: Parametric nonlinear optics with layered materials and related heterostructures. Laser Photonics Rev. 16, 2100726 (2022). https://doi.org/10.1002/lpor.202100726
  36. [36]Chang, D.E., Vuletić, V., Lukin, M.D.: Quantum nonlinear optics — photon by photon. Nat. Photonics 8, 685–694 (2014). https://doi.org/10.1038/nphoton.2014.192
  37. [37]Elshaari, A.W., Pernice, W., Srinivasan, K., Benson, O., Zwiller, V.: Hybrid integrated quantum photonic circuits. Nat. Photonics 14, 285–298 (2020). https://doi.org/10.1038/s41566-020-0609-x
  38. [38]Li, W., Wang, S., Peng, P., Han, H., Wang, X., Ma, J., Luo, J., Liu, J.-M., Li, J.-F., Nan, C.-W., Li, Q.: Terahertz excitation of collective dynamics of polar skyrmions over a broad temperature range. Nat. Phys. (2025). https://doi.org/10.1038/s41567-025-03056-8
  39. [39]Damodaran, A.R., Clarkson, J.D., Hong, Z., Liu, H., Yadav, A.K., Nelson, C.T., Hsu, S.-L., McCarter, M.R., Park, K.-D., Kravtsov, V., Farhan, A., Dong, Y., Cai, Z., Zhou, H., Aguado-Puente, P., García-Fernández, P., Íñiguez, J., Junquera, J., Scholl, A., Raschke, M.B., Chen, L.-Q., Fong, D.D., Ramesh, R., Martin, L.W.: Phase coexistence and electric-field control of toroidal order in oxide superlattices. Nature Mater. 16, 1003–1009 (2017). https://doi.org/10.1038/nmat4951
  40. [40]Lopez, D.A.B., Gyan, D.S., Das, S., Fechner, M., Li, X., Martin, L.W., Ramesh, R., Gopalan, V., Hu, W.: Phononic excitation of polar collective modes. Submitted (2025)
  41. [41]Disa, A.S., Nova, T.F., Cavalleri, A.: Engineering crystal structures with light. Nat. Phys. 17, 1087–1092 (2021). https://doi.org/10.1038/s41567-021-01366-1
  42. [42]Subedi, A.: Proposal for ultrafast switching of ferroelectrics using midinfrared pulses. Phys. Rev. B 92, 214303 (2015). https://doi.org/10.1103/PhysRevB.92.214303
  43. [43]Mankowsky, R., von Hoegen, A., Först, M., Cavalleri, A.: Ultrafast reversal of the ferroelectric polarization. Phys. Rev. Lett. 118, 197601 (2017). https://doi.org/10.1103/PhysRevLett.118.197601
  44. [44]Grishunin, K.A., Ilyin, N.A., Sherstyuk, N.E., Mishina, E.D., Kimel, A., Mukhortov, V.M., Ovchinnikov, A.V., Chefonov, O.V., Agranat, M.B.: THz Electric Field-Induced Second Harmonic Generation in Inorganic Ferroelectric. Sci. Rep. 7, 687 (2017). https://doi.org/10.1038/s41598-017-00704-9
  45. [45]Zhang, X., Carbin, T., Culver, A.B., Du, K., Wang, K., Cheong, S.-W., Roy, R., Kogar, A.: Light-induced electronic polarization in antiferromagnetic Cr₂O₃. Nat. Mater. 23, 790–795 (2024). https://doi.org/10.1038/s41563-024-01852-w
  46. [46]Freire, J.D., Katiyar, R.S.: Lattice dynamics of crystals with tetragonal BaTiO₃ structure. Phys. Rev. B 37, 2074–2085 (1988). https://doi.org/10.1103/PhysRevB.37.2074
  47. [47]Luk, T.S., de Ceglia, D., Liu, S., Keeler, G.A., Prasankumar, R.P., Vincenti, M.A., Scalora, M., Sinclair, M.B., Campione, S.: Enhanced third harmonic generation from the epsilon-near-zero modes of ultrathin films. Appl. Phys. Lett. 106, 151103 (2015). https://doi.org/10.1063/1.4917457
  48. [48]Reshef, O., Leon, I.D., Alam, M.Z., Boyd, R.W.: Nonlinear optical effects in epsilon-near-zero media. Nat. Rev. Mater. 4, 535–551 (2019). https://doi.org/10.1038/s41578-019-0120-5
  49. [49]Kinsey, N., DeVault, C., Boltasseva, A., Shalaev, V.M.: Near-zero-index materials for photonics. Nat. Rev. Mater. 4, 742–760 (2019). https://doi.org/10.1038/s41578-023-00598-2
  50. [50]Lopez, D.A.B., Juraschek, D.M., Fechner, M., Xu, X., Cheong, S.-W., Hu, W.: Ultrafast simultaneous manipulation of multiple ferroic orders through nonlinear phonon excitation. npj quantum materials 10, 24 (2025). https://doi.org/10.1038/s41535-025-00738-7
  51. [51]Akamatsu, H., Yuan, Y.K., Stoica, V.A., Stone, G., Yang, T.N., Hong, Z.J., Lei, S.M., Zhu, Y., Haislmaier, R.C., Freeland, J.W., Chen, L.Q., Wen, H.D., Gopalan, V.: Light-activated gigahertz ferroelectric domain dynamics. Phys. Rev. Lett. 120, 096101 (2018). https://doi.org/10.1103/PhysRevLett.120.096101
  52. [52]Yang, T.N., Wang, B., Hu, J.M., Chen, L.Q.: Domain dynamics under ultrafast electric-field pulses. Phys. Rev. Lett. 124, 107601 (2020). https://doi.org/10.1103/PhysRevLett.124.107601
  53. [53]Backus, E.H.G., Garcia-Araez, N., Bonn, M., Bakker, H.J.: On the role of fresnel factors in sum-frequency generation spectroscopy of metal–water and metal-oxide–water interfaces. J. Phys. Chem. C 116, 23351–23361 (2012). https://doi.org/10.1021/jp306273d
  54. [54]von Hoegen, A., Mankowsky, R., Fechner, M., Först, M., Cavalleri, A.: Probing the interatomic potential of solids with strong-field nonlinear phononics. Nature 555, 79–82 (2018). https://doi.org/10.1038/nature25484
  55. [55]Chen, L.Q., Shen, J.: Applications of semi-implicit fourier-spectral method to phase field equations. Computer Physics Communications 108, 147–158 (1998). https://doi.org/10.1016/S0010-4655(97)00115-X
  56. [56]Haun, M.J., Furman, E., Jang, S.J., McKinstry, H.A., Cross, L.E.: Thermodynamic theory of PbTiO3. J. Appl. Phys. 62, 3331–3338 (1987). https://doi.org/10.1063/1.339293
  57. [57]Sheng, G., Li, Y.L., Zhang, J.X., Choudhury, S., Jia, Q.X., Gopalan, V., Schlom, D.G., Liu, Z.K., Chen, L.Q.: A modified landau-devonshire thermodynamic potential for strontium titanate. Appl. Phys. Lett. 96, 232902 (2010). https://doi.org/10.1063/1.3442915
  58. [58]Tan, C., Dong, Y., Sun, Y., Liu, C., Chen, P., Zhong, X., Zhu, R., Liu, M., Zhang, J., Wang, J., Liu, K., Bai, X., Yu, D., Ouyang, X., Wang, J., Gao, P., Luo, Z., Li, J.: Engineering polar vortex from topologically trivial domain architecture. Nat. Commun. 12, 4620 (2021). https://doi.org/10.1038/s41467-021-24922-y

Paper details

vlnc:W_y8x8dv7u
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 1

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents